JP7599018B2 - Rainwater infiltration rate estimation device, rainwater infiltration rate estimation method, and program - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act June 20, 2020 In the publication “Environmental Purification Technology July/August 2020 Issue, Volume 19, No. 4, Volume 156, Pages 44-48” public

Article 30, paragraph 2 of the Patent Act applied July 27, 2020 Disclosure in the publication "Proceedings of the 57th Sewerage Research Presentation Conference, pages 652-654"

The present invention relates to a rainwater infiltration rate estimation device, a rainwater infiltration rate estimation method, and a program that can be applied to a system for narrowing down the area where rainwater infiltration occurs in sewerage pipes.

In a separate sewer system where the storm drain pipes for draining rainwater and the sewage pipes for draining sewage to a sewage (sewage) treatment plant are separated, if rainwater infiltrates into the sewage pipes during rainy weather (rainy weather infiltration), the amount of water flowing into the sewerage facility increases, causing problems such as overflow from the sewage pipes and increased burden on the treatment facility. These problems are strongly required as they can cause an increase in sewage treatment costs, deterioration of the water quality of public water areas, deterioration of public health due to overflow, and flood damage. In the future, it is expected that the amount of rainy weather infiltration will increase due to the deterioration of the sewage pipes. In order to efficiently reduce rainy weather infiltration, it is necessary to implement measures after narrowing down the generation area from the entire target area of the business entity that manages the sewerage system. On the other hand, since field surveys to narrow down the generation area are too expensive, a low-cost and highly accurate method of narrowing down the generation area is required.

Patent Document 1 describes an unknown water occurrence distribution estimation device that estimates the occurrence distribution of unknown water flowing into a sewerage system from a target area, and obtains the unknown water occurrence distribution in each district by performing pattern matching analysis. However, no specific method has been provided so far for dividing a target area into multiple areas (meshes) and estimating the infiltration rate during rainy weather using a machine learning model.

Patent No. 3857670

In view of the above, the present invention aims to provide a rain-time water infiltration rate estimation device, a rain-time water infiltration rate estimation method, and a program that estimates a rain-time water infiltration rate, which corresponds to the ratio of the amount of water infiltration during rainfall to the amount of rainfall, defined for each of a plurality of areas in a target area divided into a plurality of areas, and that performs estimation using an estimation model that has been subjected to machine learning.

In order to solve the above problem, the present invention provides a rainwater infiltration rate estimation device that estimates a rainwater infiltration rate, which is defined for each of a target area divided into a plurality of areas and corresponds to the ratio of the amount of infiltration water during rainfall to the amount of rainfall, and the device includes a variable data acquisition unit that acquires data on variables defined for each of the areas, the variables including the inflow target rainfall, which is a function of the amount of precipitation in each of the areas, and a rainwater infiltration rate estimation unit that estimates the rainwater infiltration rate for each of the areas using the data on the variables for each of the areas acquired by the variable data acquisition unit and an estimation model that estimates the rainwater infiltration rate from the variables, the estimation model being trained by machine learning using past data on the rainwater infiltration rate and the variables as training data.

The estimation model may be a pattern-specific estimation model corresponding to the rainfall pattern of the target rainfall, among two or more separate pattern-specific estimation models each of which has been machine-learned separately for two or more rainfall patterns using past data on rainfall infiltration rates and variables as training data.

The machine learning learning algorithm may be a random forest or a neural network.

The rainwater infiltration rate estimation device may further include a rainwater infiltration amount calculation unit that calculates an estimated value of the amount of rainwater infiltration in each area using data on the inflow target rainfall, which is a function of the amount of precipitation in each area acquired by the variable data acquisition unit, and an estimated value of the rainwater infiltration rate in each area obtained by estimation by the rainwater infiltration rate estimation unit.

The function of precipitation may be the inflow target rainfall calculated using the precipitation and area in each region.

The present invention also provides a sewage-related facility inflow estimation device, which is any of the above-mentioned devices for estimating water infiltration rates during rainy weather, further comprising: a rainy weather infiltration rate storage unit that stores from moment to moment an estimated value of the rainy weather infiltration rate for each area obtained by estimation by the rainy weather infiltration rate estimation unit in association with date and time; and a sewage-related facility inflow estimation unit that estimates the inflow volume of inflow water flowing into a sewage-related facility during rainy weather at a future time point that is a certain time later than a reference time point, and which estimates the inflow volume at a later time point using the predicted precipitation or actual precipitation for each area at a date and time determined for each area in accordance with the flow arrival time from each area to the sewage-related facility, the area of each area, the infiltration rate for each area, the estimated value of the rainy weather infiltration rate for each area at a date and time determined for each area in accordance with the flow arrival time, and the sunny weather inflow volume at a date and time estimated for each area in accordance with the flow arrival time.

The present invention also provides an estimation method for estimating a water infiltration rate during rainfall, which is executed by a water infiltration rate estimation device for estimating a water infiltration rate during rainfall, which is defined for each of a target area divided into a plurality of areas and corresponds to the ratio of the amount of water infiltration during rainfall to the amount of rainfall, the method comprising: a variable data acquisition step for acquiring data on variables defined for each of the areas, the variables including the inflow target rainfall, which is a function of the amount of rainfall in each of the areas; and a water infiltration rate estimation step for estimating the water infiltration rate during rainfall for each of the areas using the data on the variables for each of the areas acquired in the variable data acquisition step and an estimation model for estimating the water infiltration rate during rainfall from the variables, which has been subjected to machine learning using past data on the water infiltration rate during rainfall and the variables as training data.

The present invention also provides a program for causing a computer to execute a rain-time water infiltration rate estimation method for estimating a rain-time water infiltration rate, which corresponds to the ratio of rain-time water infiltration rate to rainfall, defined for each of a target area divided into a plurality of areas, the program including a variable data acquisition step for acquiring data on variables defined for each of the areas, the variables including inflow target rainfall, which is a function of precipitation in each of the areas, and a rain-time water infiltration rate estimation step for estimating the rain-time water infiltration rate for each of the areas using data on variables for each of the areas acquired in the variable data acquisition step and an estimation model for estimating the rain-time water infiltration rate from the variables, which has been subjected to machine learning using past data on the rain-time water infiltration rate and the variables as training data.

According to the present invention, by utilizing a machine learning model to estimate the infiltration rate for each area in a target area, it is possible to narrow down the area where infiltration occurs during rainy weather without on-site investigation or with a minimum of on-site investigation. This makes it possible to rationally narrow down the priority areas for countermeasures from the entire target area, which was previously difficult due to the high cost. This makes it possible to efficiently carry out maintenance such as repairs and replacements of sewer pipes (sewage pipes).
In one embodiment, it is possible to predict the inflow of sewage treatment facilities, including sewage treatment plants and storm water pumping stations, during rainy weather based on the infiltration water rate for each area, real-time precipitation, and predicted precipitation.
By accurately predicting sudden changes in inflow volume, it becomes possible to control operations in advance, reducing the workload of operators and energy consumption. In addition, by predicting in advance the inflow volume that cannot be handled, time is gained to consider how to respond, making it possible to reduce damage caused by flooding or damage to facilities.

FIG. 2 is a block diagram showing the configuration of a rainwater infiltration rate estimation device. FIG. 1 is a block diagram showing the configuration of a system for narrowing down the area where water intrusion occurs during rainfall. 4 is a flowchart showing the operation flow of a learning stage executed by the rainwater infiltration rate estimation device. A conceptual diagram showing that the area for estimating water infiltration rates during rainfall is divided into multiple areas (meshes). A diagram showing the layout of sewage (wastewater) treatment plants and sewerage pipelines (wastewater pipes) in a target area divided into multiple districts. 4 is a flowchart showing the operational flow of the operation phase executed by the rainfall water infiltration rate estimation device. A diagram explaining the concept (learning stage) of random forest as an example of a learning algorithm. A diagram explaining the concept of random forest (operational stage) as an example of a learning algorithm. FIG. 1 is a diagram for explaining the concept of a neural network as an example of a learning algorithm. FIG. 1 is a diagram for explaining the concept of estimating the inflow volume to a sewage-related facility at a future point in time that is a certain time period after the current time. A screen image of the system for narrowing down the area where water infiltration occurs during rainfall (rainfall). A screen image of the system for narrowing down the areas where water intrusion occurs during rain (model creation and learning execution). A screen image of the system for narrowing down the areas where water intrusion occurs during rain (learning results). 13A and 13B are diagrams showing analysis results in an embodiment of a system for narrowing down the area where water intrusion during rain is occurring. Graph showing a comparison of the analysis results and the actual measured water infiltration rates. A table showing a comparison of the analysis results and actual measured water infiltration rates. Schematic diagram explaining a separate sewer system. A diagram explaining the time difference between rainfall and the occurrence of infiltration water during rainy weather (amount of inflow to the treatment plant and amount of rainfall per area).

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. However, it should be noted that the scope of the present invention is not limited to the following embodiment, but is defined by the description of the claims. For example, the formulas used in the following embodiment are merely examples, and the use of these formulas is not essential in carrying out the present invention. For example, the machine learning model used to estimate the infiltration rate during rainy weather may be created by creating only one model rather than creating a model for each rainfall pattern, and estimating the infiltration rate using that one model regardless of the rainfall pattern. The machine learning algorithm for generating the machine learning model may be any algorithm, not limited to the random forest and neural network described below. The selection of explanatory variables, etc. is also arbitrary. The data format of each data may be any format, such as a CSV (Comma Separated Value) format file. Each functional unit may be realized by a single hardware unit, or may be realized by two or more hardware units, or multiple functional units may be realized by one hardware unit as described below. In this specification, "water" may be pure water, or may be water containing any impurities, such as sewage such as polluted water, rainwater, etc. The amount of water is expressed in mass units (t), with 1 ton = 1 ^m3 .

Figure 17 is a diagram explaining a separate sewer system. In a separate sewer system, a sewer pipe (sewage pipe) 1000 and a storm water pipe 1100 are separated. Sewage discharged from houses, buildings, etc. is collected through the sewer pipe 1000 to a sewage (sewage) treatment plant 1001, where it is treated and then discharged into the sea or a river. On the other hand, rainwater is either discharged directly into a public water area such as the sea or a river through the storm water pipe 1100, or collected at a storm water pumping station 1101, where sand and garbage are removed and then discharged into a public water area such as the sea or a river. Here, if the sewer pipe 1000 has a defective portion (crack, break, hole, etc.) 1002 due to deterioration over time or the like, rainwater that has seeped into the ground may enter the sewer pipe 1000 through the defective portion 1002, which may increase the processing burden of the sewage treatment plant 1001, cause a deterioration in the water quality of public water areas, a deterioration in public health due to overflow, or cause problems such as flooding damage.

FIG. 18 is a diagram for explaining the time difference between rainfall and the occurrence of rainwater infiltration in a sewage treatment plant. In the graph of FIG. 17, the horizontal axis represents time, and the vertical axis represents the inflow (t) (inflow in fine weather + inflow in fine weather) from a certain area to the sewage treatment plant 1001. In the graph, "rainfall" is the average of the precipitation for each area at the corresponding time (run-off time is not taken into account), and "rainwater infiltration amount" is calculated by calculating "actual measured inflow amount - inflow amount in fine weather". "Inflow amount in fine weather" is the average of inflow amounts for three months (taking into account days of the week, etc.). The inflow amount in fine weather is an amount that can be estimated depending on the season, weather conditions, day of the week, etc., and is treated as a known amount (an estimated value has already been obtained) in the following embodiment. The inflow amount in rainwater is (the amount of) inflow water that can be calculated using the rainwater infiltration rate estimated using a machine learning model as described later. 18, there is a time difference between the time when the amount of precipitation during rainfall peaks and the time when the amount of rainwater infiltrating into the treatment plant peaks, which is related to the time (run-off time) it takes for the rainwater infiltrating through the sewer pipes leading to the sewage treatment plant 1001 to reach the sewage treatment plant 1001. The amount of infiltrating water A for each area at the relevant time of the relevant rainfall is calculated, and learning can be carried out using this as the correct answer for machine learning.
A = (rainfall in the area) / Σ _{total area} (rainfall in each area) × (amount of water infiltrating during rain)
However, in the above formula, the arrival time is taken into consideration for the "rainfall per region" (see also FIG. 18).

Division of the target area into multiple areas (meshes) FIG. 4 is a conceptual diagram showing that the target area for estimating the infiltration rate during rainy weather is divided into multiple areas (meshes), and FIG. 5 is a diagram showing the arrangement of sewage (sewage) treatment plants and sewerage pipelines (sewage pipes) in the target area divided into multiple areas. In the following, each area is assumed to be a square area of 50 m on each side (FIG. 4 is a conceptual diagram for explanation, and the accuracy of the shape of the area is not taken into consideration), but the shape and size of the area can be determined arbitrarily. The "target area" is any area that is the subject of the estimation of the infiltration rate during rainy weather, and by dividing the target area into multiple areas, the target area can be captured as a collection of multiple areas. In the example of FIG. 5, the target area is given as a collection of 8×11=88 rectangular areas (square areas with a side length of 50 m), such as areas 49 and 51. Although the target area is shown as a rectangle in Figs. 4 and 5, the target area is usually the entire area flowing into the treatment plant or pump station shown in 48, and has an irregular shape.

で定義される「雨天時浸入水率」のことである（上記（１）式の雨天時浸入水率に１００を乗じることにより百分率で表すこともある）。ただし、既に述べたとおり「ｔ」は重さの単位のトンである（１トン＝１０００ｋｇ）。（１）式中、「雨天時浸入水量」とは、上記或る区域（当該区域）に降った雨水のうち、下水道管に流れ込む雨水の量（重さ。単位はトン）である。（１）式中の「降水量（ｔ）」は、上記或る区域に降った雨水の量（重さ。単位はトン）である。なお、以下において「流入対象雨量」とは、当該区域における降水量（ｍｍ）や、当該区域の面積（ｍ²）、そして後述のとおり当該区域の用途（建物、道路、河川・池、その他）に応じて所定のルールで算出される当該区域の浸透率（０以上、１以下の値）を用いて、個別の区域ごとに以下の（２）式

で定義される（雨水の密度を、１０００ｋｇ／ｍ³と仮定した。雨水、或いは下水等の密度が異なる場合は、（２）式で得られた値に適宜密度を乗じる等して補正すればよい。他の式においても同様。）。なお、ここでは流入対象雨量をｔ（トン）の単位で示しているが、（２）式中、区域の降水量（ｍｍ）は現在の例においては１時間あたりの降水量（１時間降った雨水がどこにも流れ出すことなく、その場所にたまった場合の水の深さ）と同じであるから、流入対象雨量（ｔ）も１時間あたりの流入対象雨量を意味することに留意する。同様に、（１）式の浸入水量（ｔ）も、現在の例においては１時間あたりの浸入水量を意味する。後述の各式における（処理場の）流入対象雨量、浸入水量、下水関連施設への流入量等も同様に時間あたりの量を示すことに留意する。また区域の降水量とは、例えば一辺が５０ｍの区域内における１時間あたりの降水量（ｍｍ）であるが、降雨データを後述のＸＲＡＩＮから取得する場合は、ＸＲＡＩＮの降雨データが２５０ｍ×２５０ｍの区域単位で与えられるため、５０ｍ×５０ｍの区域２５個において１つの降水量（ｍｍ）が対応することとなる（２５個の区域に同じ値の降水量を割り当てるか、または隣接するＸＲＡＩＮの区域における降水量との差を考慮して傾斜した降水量を割り付ける）。The " rainwater infiltration rate " defined individually for each area is a quantity corresponding to the ratio of the amount of water infiltration during rain to the amount of rainfall. The "rainwater infiltration rate" in a certain area is expressed by the following formula (1):

(The rainwater infiltration rate in equation (1) above may be multiplied by 100 to express it as a percentage.) However, as already mentioned, "t" is the unit of weight, ton (1 ton = 1000 kg). In equation (1), "rainwater infiltration amount" refers to the amount of rainwater (weight, unit is tons) that flows into the sewer pipes from the rainwater that falls in the certain area (the area in question). "Precipitation (t)" in equation (1) refers to the amount of rainwater that falls in the certain area (weight, unit is tons). In the following, "inflow target rainfall" refers to the amount of rainwater that flows into the sewer pipes (weight, unit is tons) calculated according to the amount of rainfall (mm) in the area, the area's area ( ^m2 ), and the infiltration rate (value between 0 and 1) of the area calculated according to a predetermined rule depending on the area's use (building, road, river/pond, etc.) as described later, using the following equation (2) for each individual area:

(The density of rainwater is assumed to be 1000 kg/ ^m3 . If the density of rainwater or sewage is different, the value obtained by formula (2) can be corrected by multiplying it by the appropriate density. The same applies to other formulas.) Note that the inflow target rainfall is shown here in units of t (tons), but in formula (2), the precipitation (mm) of the area is the same as the precipitation per hour in the current example (the depth of water when rainwater that has fallen for one hour does not flow out and accumulates in that place), so the inflow target rainfall (t) also means the inflow target rainfall per hour. Similarly, the inflow water volume (t) in formula (1) also means the inflow water volume per hour in the current example. Note that the inflow target rainfall (at the treatment plant), the inflow water volume, the inflow volume to sewage-related facilities, etc. in each formula described later also indicate the amount per hour. Furthermore, the precipitation amount for an area is, for example, the amount of precipitation (mm) per hour in an area with one side measuring 50 m. When rainfall data is obtained from XRAIN, which will be described later, the rainfall data from XRAIN is given in area units of 250 m x 250 m, so one precipitation amount (mm) corresponds to 25 areas of 50 m x 50 m (the same precipitation amount is assigned to the 25 areas, or a graded precipitation amount is assigned taking into account the difference in precipitation in adjacent XRAIN areas).

As described below, during the operational stage of the system for narrowing down the areas where infiltration water occurs during rainfall, the infiltration water rate estimation device during rainfall takes variables (watershed characteristics) defined individually for each area as input, and estimates the infiltration water rate as output individually for each area using a machine learning estimation model, and calculates an estimated value of the infiltration water volume for each area by multiplying the inflow target rainfall calculated for each area according to equation (2) using the precipitation value by the estimated infiltration water rate for that area (see equation (1) above).

1 is a block diagram showing the configuration of a rainwater infiltration rate estimation device. The rainwater infiltration rate estimation device 1 comprises a control unit 2, a storage unit 3, an input/output unit 4, and a communication unit 5. When input and display are performed via the communication unit described later, the input/output unit may be unnecessary.

The control unit 2 includes a processor 6 such as a CPU (Central Processing Unit) and a temporary memory 7 such as a RAM (Random Access Memory). When the processor 6 executes a variable data acquisition program stored in the memory unit 3, the control unit 2 (the processor 6 of the control unit 2, which also applies below) functions as a variable data acquisition unit 8. When the processor 6 executes a machine learning program stored in the memory unit 3, the control unit 2 functions as a machine learning unit 9. When the processor 6 executes a rainwater infiltration rate estimation program stored in the memory unit 3, the control unit 2 functions as a rainwater infiltration rate estimation unit 10. When the processor 6 executes a rainwater infiltration amount calculation program stored in the memory unit 3, the control unit 2 functions as a rainwater infiltration amount calculation unit 11. When the processor 6 executes a sewage-related facility inflow amount estimation program stored in the memory unit 3, the control unit 2 functions as a sewage-related facility inflow amount estimation unit 12. The processor 6 executes various control and display programs (including an operating system, application software for a geographic information system (GIS), driver software for various devices, etc.) stored in the memory unit 3, causing the control unit 2 to function as various control and display units 13. Any other program may be stored in the memory unit 3, and the processor 6 of the control unit 2 executes any program, allowing the control unit 2 to function as any functional unit. When there are a large number of regions, the amount of calculation increases when calculations are performed in units of time shorter than one hour, and the calculations may take too long. In that case, a specification may be adopted in which part of the program is performed by a GPU (Graphics Processing Unit) with high computing performance.

The variable data acquisition unit 8 is a functional unit that acquires variable data by acquiring data from an external server, various data stored in the memory unit 3, data from a database, and performing calculations such as calculating the target inflow rainfall. The variable data acquisition unit 8 stores (records) the acquired variable data in the memory unit 3 as a data record in the variable database 22. The variable data stored in the memory unit 3 (after being read out by the control unit 2 and appropriately stored in temporary memory 7; the same applies to the data processing described below) is used for rainwater infiltration rate estimation processing by the rainwater infiltration rate estimation unit 10, etc.

The machine learning unit 9 is a functional unit that uses the teacher data 17 (variable data of one or more types of variables that are explanatory variables, and past data as the infiltration rate data of the (rainy weather) infiltration rate that is the objective variable) stored in the memory unit 3 to create and update an estimation model (prediction model) that estimates (predicts) the infiltration rate from variables using a machine learning (supervised learning) algorithm. Here, the estimation model of the infiltration rate is a specific calculation formula, calculation method, parameter values (weighting coefficient values, etc.) used by the rainy weather infiltration rate estimation unit 10 described later to accept variable data as input data and output an estimated value of the infiltration rate as output data. The machine learning unit 9 stores data (which may include programs, etc.) representing the created estimation model of the infiltration rate in the memory unit 3 as the learned model 16.

The rainwater infiltration rate estimation unit 10 is a functional unit that uses the trained model 16 generated by the machine learning unit 9 to accept variable data as input data and output an estimated value of the infiltration rate as output data. The rainwater infiltration rate estimation unit 10 calculates an estimated value of the infiltration rate from the variable data by calculation using the trained model for each of the multiple areas that make up the target area. The rainwater infiltration rate estimation unit 10 stores the calculated estimated value of the infiltration rate in the memory unit 3 as a data record of the rainwater infiltration rate database 23. The rainwater infiltration rate estimation unit 10 repeatedly calculates the estimated value of the infiltration rate in each area at a predetermined time interval, for example, and repeatedly stores the estimated value of the infiltration rate in the rainwater infiltration rate database 23 of the memory unit 3 (sometimes referred to as the "rainwater infiltration rate storage unit") from time to time in association with the area and the date and time (date and time).

The rainwater infiltration calculation unit 11 is a functional unit that calculates an estimated value of the amount of infiltration water in each area according to the above-mentioned formulas (1) and (2). For each of the multiple areas that make up the target area, the rainwater infiltration calculation unit 11 calculates an estimated value of the amount of infiltration water in the area by multiplying the estimated value of the infiltration water rate in the area by the inflow target rainfall in the area. The rainwater infiltration calculation unit 11 stores the calculated estimated value of the amount of infiltration water in the memory unit 3 as a data record of the facility inflow amount and rainwater infiltration amount database 21. The rainwater infiltration calculation unit 11 repeatedly calculates the estimated value of the amount of infiltration water in each area at a predetermined time interval, for example, and repeatedly stores the estimated value of the amount of infiltration water in the memory unit 3 (sometimes referred to as the "rainwater infiltration storage unit") in association with the area and date and time from time to time in the facility inflow amount and rainwater infiltration amount database 21.

The sewage-related facility inflow amount estimation unit 12 is a functional unit that estimates the amount of inflow water at a certain point in time (for example, t minutes after a reference time point of time) of inflow water flowing from each area through a sewer pipe (sewage pipe) into a sewage-related facility such as a sewage treatment plant. The sewage-related facility inflow amount estimation unit 12 stores the calculated estimated inflow amount data in the memory unit 3 as a data record of the facility inflow amount, rainwater inflow amount database 21. Since it takes different flow time for the rainwater that infiltrates into the sewer pipe from each area to reach the sewage-related facility from that area, the inflow amount into the sewage-related facility is estimated by a calculation that takes this flow time into account. The specific calculation of the inflow amount will be explained in detail later.

The storage unit 3 is a storage (recording) device equipped with a hard disk drive, SSD (Solid State Drive), etc., and stores machine learning related programs 14 (including a variable data acquisition program, a machine learning program, and a rainwater infiltration rate estimation program) and various programs 15 (including a rainwater infiltration amount calculation program, a sewage-related facility inflow amount estimation program, various control and display programs, etc.) as programs executed by the processor 6 to make the control unit 2 function as various functional units as described above. In addition, the storage unit 3 stores, as the trained models 16 generated by the machine learning unit 9 as described above, M models, namely a model for rainfall pattern 1, a model for rainfall pattern 2, ..., a model for rainfall pattern M (M is any integer equal to or greater than 2), as two or more pattern-specific estimation models each of which has been machine-learned separately corresponding to two or more rainfall patterns, as will be described in detail later. The memory unit 3 further stores past variable data (explanatory variable data) and rainy weather water infiltration rate data (objective variable data) as teacher data 17 used by the machine learning unit 9 to generate and update the infiltration water rate estimation model. The memory unit 3 also stores the variable data and rainy weather water infiltration rate data as test data 18 for evaluating the performance of the infiltration water rate estimation model. The memory unit 3 also stores various other data 19 such as sunny weather average inflow data. The memory unit 3 further stores the following various databases.
(1) Rainfall Information Database 20
A database that stores the amount of precipitation at each date and time for each area included in a target region in the form of records whose data items include "area identifier,""date and time (can also be time zone; similarly below)," and "amount of precipitation (mm) per unit time (for example, one hour, but is not limited to this and can also be a shorter unit time) (hereinafter referred to simply as "precipitation amount")."
(2) Database of facility inflow and rainwater infiltration volume 21
A database (including at least two tables) that stores the amount of water infiltration during rain for each area at each date and time calculated by the rainfall water infiltration amount calculation unit 11 in the form of a record including the data items of ``area identifier,'' ``date and time,'' and ``amount of water infiltration,'' and stores the amount of inflow into sewage-related facilities at each date and time calculated by the sewage-related facility inflow amount estimation unit 12 in the form of a record including the data items of ``date and time'' and ``inflow amount.''
(3) Variable Database 22
A database containing variables for each area at each date and time acquired by the variable data acquisition unit 8 in the form of a record containing "area identifier", "date and time", "variable 1", "variable 2", ... "variable L" as data items. L is the number of variables as explanatory variables and is an arbitrary integer of 1 or more. In addition, as described later, in the case where only the inflow target rainfall among the variables changes from moment to moment and the other variables can be considered not to change over time (in the short term), a database storing "area identifier", "date and time", and "inflow target rainfall (variable L)" in the form of a record containing the data items may be prepared as a first variable database, and a database storing "area identifier", "variable 1", "variable 2", ... "variable L-1" in the form of a record may be prepared as a second variable database.
(4) Rainfall Water Infiltration Rate Database 23
A database that stores the rainwater infiltration rate for each area at each date and time calculated by the rainwater infiltration rate estimation unit 10 in the form of records that include the data items ``area identifier,'' ``date and time,'' and ``infiltration rate.''
(5) Geographical Information Database 24
A database that stores geographic information for each area acquired by the variable data acquisition unit 8 from an external server machine (geographic information server) 31 or the like (see FIG. 2 described later) in the form of records including data items such as "area identifier,""geographic information 1,""geographic information 2," ... and "geographic information P." Note that P is the number of pieces of geographic information given to a certain area and is an arbitrary integer equal to or greater than 1.

The input/output unit 4 includes input devices such as a keyboard 25 and a mouse 26 for the system administrator to input commands and data into the rainwater infiltration rate estimation device 1, and output devices such as a display device 27 (a liquid crystal display device, an organic electroluminescence (organic EL) display device, etc.) for displaying information such as images. The input/output unit may not be necessary if input and display are performed via the communication unit described below. The communication unit 5 includes a communication interface 28 and a communication circuit 29 for communication between the rainwater infiltration rate estimation device 1 and external server machines, client machines, various devices, etc.

2 is a block diagram showing the configuration of the system for narrowing down the area where water infiltration occurs during rainy weather. The system for narrowing down the area where water infiltration occurs during rainy weather is configured by the rainy weather infiltration rate estimation device 1, an external server machine (geographic information server) 31, a sewage-related facility 32, a precipitation measurement system 33, and a client machine 34, which are linked together via a communication line 30 such as the Internet.

The external server machine (geographic information server) 31 is, in one example, a server machine that stores geographic information used when the control unit 2 of the rainwater infiltration rate estimation device 1 uses a geographic information system (GIS), and has a control unit, memory unit, input/output unit, and communication unit, just like the rainwater infiltration rate estimation device 1. The memory unit of the external server machine 31 stores map data 35, land use data 36, ground rain gauge position data 37, and other geographic data 38, and transmits these data to the rainwater infiltration rate estimation device 1 in response to a request from the rainwater infiltration rate estimation device 1. The geographic information server 31 may be composed of multiple independent servers, and various data may be distributed and stored in multiple servers, for example, with the map data 35 stored in a first geographic information server and the land use data 36 stored in a second geographic information server.

The sewage-related facility 32 is a facility such as a sewage (sewage) treatment plant, a pumping station, etc., and will be referred to as a sewage (sewage) treatment plant hereinafter. The inflow flow rate is measured at any time at the sewage treatment plant 32, and the memory unit of a computer installed at the sewage treatment plant 32 stores inflow flow rate actual measurement data 39 that associates the actual value of the inflow flow rate with the measurement date and time. The computer installed at the sewage treatment plant 32 transmits the inflow flow rate actual measurement data 39 to the precipitation-time water infiltration rate estimation device 1 in response to a request from the precipitation-time water infiltration rate estimation device 1.

The precipitation measurement system 33 is a system including a weather radar, a rain gauge, etc. for measuring the amount of precipitation that changes depending on the location and date and time, and in one example is configured using XRAIN (extended radar information network). The memory unit of the computer included in the precipitation measurement system 33 stores rainfall information data and forecast rainfall information data 40 including actual and forecast values of precipitation for each location at each date and time. The computer included in the precipitation measurement system 33 transmits the rainfall information data and forecast rainfall information data 40 to the rainfall infiltration rate estimation device 1.

The client machine 34 is a device such as a computer or a smartphone, and includes a control unit 41, a memory unit 42, an input/output unit (display unit) 43, and a communication unit 44. A system administrator or other user can input commands to the rainwater infiltration rate estimation device 1 from the input/output unit (display unit) 43 of the client machine 34 to cause the rainwater infiltration rate estimation device 1 to execute each of the above-mentioned functions, and display on the display device of the client machine 34 the estimated values of the infiltration water rate and infiltration water volume obtained as a result of the execution, or an image showing the distribution of these estimated values on a map, or an image showing the estimated value of the inflow volume into a sewerage-related facility (time series display is also possible).

Operation of the device for estimating a rate of infiltration of water during rain and the system for narrowing down an area where infiltration of water during rain will now be described with reference to Figures 3 to 10 as well.

1. Operation in the Learning Phase Fig. 3 is a flow chart showing the operation flow in the learning phase executed by the rainwater infiltration rate estimating device.

Generation of Area (Mesh) Information First, in step S301, a map is read and area (mesh) information is generated. In response to the rainwater infiltration rate estimation device 1 receiving an instruction from a user such as an administrator via the input/output unit 4 or the input/output unit (display unit) 43, the processor 6 of the rainwater infiltration rate estimation device 1 executes a variable data acquisition program, and the variable data acquisition unit 8 requests the map data 35, land use data 36, ground rain gauge position data 37, and other geographical data 38 from the geographical information server 31. The geographical information server 31 that has received the request transmits the map data 35, land use data 36, ground rain gauge position data 37, and other geographical data 38 to the rainwater infiltration rate estimation device 1. The variable data acquisition unit 8 generates mesh information within the target area based on information indicating the target area (which area in the area indicated by the map data 35 is to be the target) and the size of the area (mesh) (how finely the target area is to be divided), which are designated in advance by a user such as an administrator through the input/output unit 4 or the input/output unit (display unit) 43. In one example, the mesh information is information consisting of the coordinates of the reference position of the mesh (the center point, the upper left corner, etc.) and the size of the mesh (the length of one side, or the length of each of the vertical and horizontal directions, etc.), and the data of the mesh information is stored in the geographic information database 24 of the storage unit 3. When information from a rain gauge installed within the target area is used as the rainfall information, the variable data acquisition unit 8 selects the rain gauge of the analysis unit and stores the information in the geographic information database 24.

Flow time setting Next, in step S302, the flow time in each area is set. The "flow time" is the time required for infiltration water to flow down the sewer pipe from each area to reach the destination (e.g., sewage treatment plant), and is defined separately for each area. One "flow time" is defined for each area as the average value of the flow time in one area. The flow time in each area is determined in advance by calculation, measurement survey, etc. (in one example, the flow rate is calculated for each pipe using the Manning formula from the pipe specifications in the sewer ledger, the flow time from the pipe to the treatment plant is obtained, and the average value in each area is calculated and assigned to each area.) and is stored in the memory unit 3 as part of the various data 19.

In step S303 of setting the terminal inflow and various variables , the variable data acquisition unit 8 sets various variables for each mesh. One of the variables set for each area is "land use (permeability) information." Land use (permeability) information is information (a value between 0 and 1) indicating the average permeability (how easily rainwater permeates into the land) within one mesh, and is set by calculating a weighted average of the permeability values given for each use (obtained by adding up for all uses the "product of the proportion (between 0 and 1) of the area of the mesh used for a certain use and the permeability for that use") according to how much area of the land in the mesh is used for what purpose.

となる。変数データ取得部８は、このようなメッシュの浸透率を対象地域内の全てのメッシュについて計算し、各々のメッシュの土地利用（浸透率）情報として記憶部３の変数データベース２２に格納させる。なお、各メッシュにおける、「当該メッシュの土地のうち、どれだけの面積がどのような用途に利用されているか」の情報、及び用途ごとの浸透率の値は、管理者等のユーザが入出力部４或いは入出力部（表示部）４３を介して入力すること等により地理情報データベース２４に予め格納されていることとしてもよいし、地図データ３５に含まれる衛星写真を変数データ取得部８が分析することで「当該メッシュの土地のうち、どれだけの面積がどのような用途に利用されているか」を特定して、その情報を地理情報データベース２４に格納する等してもよい。その他の例として、各区域の浸透率は、予め設定せずに学習させる方法によって決めてもよい。 In one example, the penetration rate for each application is as follows:
Building… 0.0
Road... 0.1
Rivers and ponds: 1.0
Others: 0.8
For example, if 40% of the area of a mesh is occupied by buildings, 20% by roads, 10% by rivers or ponds, and 30% is used for other purposes, the (average) infiltration rate of the area is

The variable data acquisition unit 8 calculates the infiltration rate of such a mesh for all meshes in the target area, and stores the infiltration rate in the variable database 22 of the storage unit 3 as land use (infiltration rate) information for each mesh. Note that the information on "how much area of the land in the mesh is used for what purpose" and the infiltration rate value for each purpose for each mesh may be stored in advance in the geographic information database 24 by a user such as an administrator inputting the information through the input/output unit 4 or the input/output unit (display unit) 43, or the variable data acquisition unit 8 may analyze a satellite photo included in the map data 35 to identify "how much area of the land in the mesh is used for what purpose" and store the information in the geographic information database 24. As another example, the infiltration rate of each area may be determined by a method of learning without being set in advance.

（表１） In addition, the variable data acquisition unit 8 acquires and sets the values of the basin characteristics used as the values of the explanatory variables in the teacher data of the machine learning for each area, and stores them in the storage unit 3 as the variable data (explanatory variable data) of the teacher data 17. In addition, among the variable data, the data values other than the inflow target rainfall (or rainfall) can be regarded as constants that do not change at least on a short time scale, and can also be used as explanatory variable data during operation. Therefore, the variable data acquisition unit 8 also stores, among the variable data indicating the acquired and set values of the basin characteristics, at least the data other than the inflow target rainfall (or rainfall) in the variable database 22. A list of explanatory variables for machine learning in this embodiment is as shown in Table 1 below. Note that, among the explanatory variables, for those related to sewage pipes such as "sewage pipe installation year", for an area where no sewage pipe exists, such as area 51, information on the inflow pipe of the area is assigned (the installation year of the sewage pipe 52 is used as the "sewage pipe installation year" of area 51).

(Table 1)

The residential area (density) is the ratio of residential area to the area of each district (in this embodiment, each district is a square with sides of 50 m, so the area is 2500 ^m2 ), and for example, if the residential area in a certain district is 1000 ^m2 , the residential area (density) value is 1000/2500 = 0.4. The variable data acquisition unit 8 uses the geographic information stored in the geographic information database 24 (other geographic data 38 includes information indicating which areas in the map data are residential. In one example, buildings may be considered as residential areas) to identify the residential area in each district and calculate the value of the residential area (density) as an explanatory variable.

As already explained, land use (permeability) is set for each area by the variable data acquisition unit 8 in step S302. The variable data acquisition unit 8 determines the land use (permeability) value (0 to 1) set for each area as the land use (permeability) value as an explanatory variable.

The sewer pipe installation year is the year in which the sewer pipes existing in each area (underground) were installed, but if there are no sewer pipes in the area, it may be the year in which the inflow pipeline of the area was installed. In one example, in a target area divided into multiple areas (8 x 11 = 88 areas) as shown in Figure 5, a sewer pipe 50 exists in area 49 (underground), so the value of the sewer pipe installation year as an explanatory variable in area 49 is the year in which the sewer pipe 50 was installed (if it is 1980, it is "1980"). Since there are no sewer pipes in area 51, it does not need to be the subject of machine learning, but since there is a sewer pipe 52 as the closest inflow pipeline to area 51, the value of the sewer pipe installation year as an explanatory variable in area 51 may be the year in which the sewer pipe 52 was installed (if it is 2000, it is "2000"). The absolute value of the installation year is not important, and the year the first pipe was installed may be set to 1, or the current or future point in time may be set to 0 to indicate how many years ago it was. Information indicating which sewage pipes exist in each area, or which sewage pipes correspond to each area as inflow pipelines, and the value of the installation year of each sewage pipe, in one example, is stored in advance as part of the various data 19 using information such as the location on a map, pipe type, and installation year recorded in GIS data owned by the sewerage administrator (the correspondence between areas and sewage pipes may be determined by the variable data acquisition unit 8 using geographic information stored in the geographic information database 24).

The number of sewage manholes is the number of sewage manholes (manholes) that exist in each area and that connect to underground sewage pipes. Information indicating how many sewage manholes exist in each area may be stored in advance as part of various data 19, for example, using information such as the location on the map, pipe type, and installation year recorded in the GIS data owned by the sewerage administrator, or may be determined by the variable data acquisition unit 8 using geographic information stored in the geographic information database 24. If there are three sewage manholes in a certain area, the value of the number of sewage manholes as an explanatory variable for that area will be "3" (if there is no sewage manhole in the area, the value of the number of sewage manholes will be "0").

The number of sewage manholes is the number of sewage manholes in each area, and can be determined by the number of sewage junction pipes that connect the sewage manholes to sewer (sewage) pipes in the area. In one example, information such as the location on the map, pipe type, and installation year recorded in GIS data owned by the sewerage administrator may be used and may be stored in advance as part of various data 19, or the variable data acquisition unit 8 may determine the number of sewage manholes using geographic information stored in the geographic information database 24. If there are five sewage manholes in a certain area, the value of the number of sewage manholes as an explanatory variable for that area will be "5" (if there are no sewage manholes in the area, the value of the number of sewage manholes will be "0").

Sewer pipe length is the total length (m) of sewer pipes existing in each area (underground). In one example, information such as the location on the map, pipe type, and installation year recorded in GIS data owned by the sewerage administrator may be used and may be stored in advance as part of various data 19, or the variable data acquisition unit 8 may determine it using geographic information stored in the geographic information database 24. If there is a 70m long sewer pipe in a certain area, the value of the sewer pipe length as an explanatory variable for that area will be "70" (if there is no sewer pipe in the area, the value of the sewer pipe length will be "0").

The length of the earthen pipe is the total length (m) of the earthen pipe in each area (underground). In one example, information such as the location on the map, the type of pipe, and the year of construction recorded in the GIS data owned by the sewerage manager may be used and may be stored in advance as part of the various data 19, or the variable data acquisition unit 8 may determine the length of the earthen pipe using geographic information stored in the geographic information database 24. If there are 30 m of earthen pipes in a certain area, the value of the earthen pipe extension as an explanatory variable in that area will be "30" (if there are no earthen pipes in the area, the value of the earthen pipe extension will be "0"). The value of the earthen pipe extension as an explanatory variable may be the sum of the total length (m) of the earthen pipes in each area (underground) and the total length (m) of the Hume pipes.

Stormwater pipe length is the total length (m) of stormwater pipes existing in each area (underground). In one example, information such as the location on the map, pipe type, and installation year recorded in GIS data owned by the sewerage administrator may be used and may be stored in advance as part of various data 19, or the variable data acquisition unit 8 may determine it using geographic information stored in the geographic information database 24. If there is a 50m stormwater pipe in a certain area, the value of the stormwater pipe length as an explanatory variable for that area will be "50" (if there is no stormwater pipe in the area, the value of the stormwater pipe length will be "0").

Calculating the inflow volume on sunny days Next, in step S304, the various control and display units 13 calculate the inflow volume on sunny days (average). The "average inflow volume on sunny days" is the average inflow volume on sunny days to the sewage treatment plant that can be estimated according to the season, weather conditions, day of the week, etc., and the estimated value is calculated by the sewage-related facility inflow volume estimation unit 12 and stored in the memory unit 3 as part of the various data 19. In this example, the average inflow volume on sunny days is the three-month average inflow volume for each day of the week on sunny days (for example, the average inflow volume on sunny days on Mondays is the average value of the actual inflow volume measured on Mondays on sunny days over a three-month period, and the three-month average value of the actual measured value for the same day is calculated similarly for other days of the week).

In the step S305 of acquiring past rainfall data (for the target rainfall pattern) , the variable data acquiring unit 8 acquires past rainfall data. Specifically, the rainfall data is distributed in real time from the device of the precipitation measurement system 33. This distributed data is used for predicting the inflow amount. The acquisition of past data used for narrowing down the rainwater infiltration occurrence area is performed upon receiving an order from the precipitation measurement system 33. After placing an order, the data is downloaded or received via an external storage medium. The variable data acquiring unit 8 acquires past rainfall data from the computer of the precipitation measurement system 33 and stores it in the rainfall information database 20. The computer of the precipitation measurement system 33 transmits rainfall information data and forecast rainfall information data 40 indicating measured values of precipitation at various past dates and times at each point in the target area, and forecast values of precipitation at various future dates and times at each point as necessary, to the rainwater infiltration rate estimation device 1. The variable data acquiring unit 8 stores the acquired rainfall information data and forecast rainfall information data in the rainfall information database 20.

In this embodiment, since an estimation model of the rainwater infiltration rate is generated separately for each rainfall pattern, the rainfall information database 20 may store rainfall information data after classifying the data according to the rainfall pattern in advance (for example, by including an identifier indicating the rainfall pattern in the data item).
Long-term rainfall: When rainfall continues for a certain period of time, such as 5 hours or more, and for a certain rainfall intensity or more,
Localized concentrated rainfall (forward concentration): When, for example, 70% or more of the total precipitation (accumulated value of precipitation over time) falls in the first half of the rainfall period,
Localized concentrated rainfall (rear concentration): When, for example, 70% or more of the total precipitation falls in the latter half of the rainfall period,
Heavy rain: For example, total precipitation of 90 mm or more,
Heavy rain: For example, total precipitation between 50 mm and 90 mm.
Moderate rain: For example, total precipitation is between 10 mm and 50 mm.
Light rain: For example, total precipitation is between 5mm and 10mm.
Light rain: for example, total precipitation less than 5 mm,
Other: If it does not fall into any of the above patterns,
The rainfall patterns are defined as above (if a certain rainfall corresponds to two or more rainfall patterns, it can be considered to correspond only to a specific rainfall pattern with the highest priority predefined for each rainfall pattern, or it can be associated with all the corresponding patterns), and the rainfall information data can be stored in the rainfall information database 20 in association with an identifier for identifying the rainfall pattern. The rainfall patterns can be defined arbitrarily (for example, heavy rain, strong rain, and moderate rain can be divided into two patterns depending on whether the maximum instantaneous rainfall is 10 mm or more or less than 10 mm), and in one example, a user such as an administrator inputs data via the input/output unit 4 or the input/output unit (display unit) 43, and the data is stored in advance as part of the various data 19. However, it is not essential to generate a model for each rainfall pattern separately.

In step S306 of calculating the amount of infiltration water during rain for the target rainfall , the infiltration water amount calculation unit 11 calculates the amount of infiltration water during rain as part of the teacher data 17. As shown in Fig. 18, the amount of infiltration water during rain is calculated by subtracting the inflow amount during sunny weather (t) from the inflow amount (t) to the treatment plant. The amount of infiltration water during rain here is used as "inflow amount during rain - average inflow amount during sunny weather" in formula (7) described later.

In step S307 of calculating the inflow target rainfall for each area , the variable data acquisition unit 8 calculates the "inflow target rainfall" in Table 1 for each area and stores it in the storage unit 3 as variable data (explanatory variable data) of the teacher data 17. Among the explanatory variables in Table 1, the inflow target rainfall is the inflow target rainfall (t) defined by the above formula (2) in each area, which is calculated by the variable data acquisition unit 8 in step S307 and stored in the storage unit 3 as variable data (explanatory variable data) of the teacher data 17. Unlike other explanatory variables, the inflow target rainfall is an amount that changes on a short time scale depending on the date and time, and the infiltration water rate as the objective variable described later is also an amount that changes on a short time scale depending on the date and time.

In this embodiment, the estimation model for water infiltration rate during rainy weather is an estimation model that accepts "a set of explanatory variable values in a certain area" as input data and outputs "an estimated value of the objective variable (water infiltration rate) in that area" as output data.

より３．２（ｔ）と算出される。この場合の説明変数としての「流入対象雨量」の値は３．２である。変数データ取得部８は、このようにして、降雨情報データベース２０に格納された各区域、各日時の降水量データ、各区域の面積データ（地理情報データベース２４に格納されているとする）、教師データ１７として格納されている各区域の浸透率を用いて、各区域、各日時における流入対象雨量の値を算出して教師データ１７の変数データとして格納する。 The variable data acquisition unit 8 calculates the inflow target rainfall (t) at each date and time (time zone) in each zone included in the target zone using the precipitation (mm) data at each date and time (time zone). In one example, in the target zone shown in FIG. 5, if rainfall occurs from 19:00 to 22:00 on May 1, 2021, and the precipitation in zone 51 is 2 mm per hour at 21:00 on that day, the inflow target rainfall in zone 51 at 21:00 on May 1, 2021 is calculated by the following formula (4) with the infiltration rate of zone 51 ("land use (infiltration rate)" among the explanatory variables) set to 0.36.

The calculated value is 3.2 (t). In this case, the value of the "inflow target rainfall" as an explanatory variable is 3.2. In this way, the variable data acquisition unit 8 calculates the value of the inflow target rainfall for each area and each date and time using the precipitation data for each area and each date and time stored in the rainfall information database 20, the area data for each area (assumed to be stored in the geographic information database 24), and the infiltration rate for each area stored as the teacher data 17, and stores the calculated value as variable data of the teacher data 17.

As the verification results described below show, the feature that is particularly highly related to the infiltration rate, which is the objective variable, is "rainfall" (precipitation). However, as shown in the above formula (2), in the calculation of the inflow target rainfall, variables other than "precipitation" can be considered as constants that do not change at least on a short time scale (in formula (2), "area of the area" is a constant, and although "infiltration rate" varies depending on the area, as the verification results described below show, its degree of correlation with the infiltration rate is very small compared to rainfall). Therefore, the time change of the inflow target rainfall as a variable can be considered to be substantially due to the time change of precipitation. Therefore, among the explanatory variables listed in Table 1, the explanatory variable that is particularly highly related to the infiltration rate, which is the objective variable, (in the example of random forest, the importance of the feature) is considered to be the inflow target rainfall. However, instead of "inflow target rainfall", for example, precipitation (mm) per hour (or the average value of precipitation (mm) for a specified time) may be used as the explanatory variable.

In machine learning, particularly when generating a separate estimation model for each rainfall pattern, the teacher data 17 is also stored separately for each rainfall pattern in the storage unit 3. As described above, in the teacher data 17, the sets of variable data and rainwater infiltration rate data are stored in a format associated with information indicating the area and date and time (time zone), but in one example, the sets of variable data and rainwater infiltration rate data are stored in a format further associated with information indicating the rainfall pattern. That is, the data stored as the teacher data 17 is
Teacher data for rainfall pattern 1 Teacher data for rainfall pattern 2 …
The training data for rainfall pattern M is classified by pattern (M is any integer equal to or greater than 2), and the training data for each rainfall pattern is stored as a set of variable data and rainwater infiltration rate data associated with information indicating the area and date and time (time zone) (it is possible to read out a set of variable data and rainwater infiltration rate data by specifying the rainfall pattern, area, and date and time (time zone)). However, it is not essential to create a separate model for each rainfall pattern, and it is sufficient to create only one model.

（表２）
上述の教師データ１７における雨天時浸入水率の値は、ステップＳ３０６において変数データ取得部８により、以下の（５）～（８）式に従って算出される。In step S308 of provisionally setting the infiltration water rate for each area , the variable data acquisition unit 8 provisionally sets the infiltration water rate for each area. As described above, the objective variable in the estimation model of this embodiment is the rainwater infiltration rate.

(Table 2)
The value of the rainwater infiltration rate in the above-mentioned teacher data 17 is calculated by the variable data acquisition unit 8 in step S306 according to the following equations (5) to (8).

処理場の流入対象雨量（ｔ）は雨天時浸入水率を推定する単位時間における量であり例えば１時間ごとの推定では１時間積算値である。ただし、（５）式中、ｋは区域の識別子であり、ｎは対象地域に含まれる区域の総数である。すなわち（５）式において、Σ記号内の式は各区域について計算される量である。また（５）式中、「流達時間前の降水量（ｍｍ）」とは、例えば着目している区域の流達時間が「ａ分」であった場合、算出すべき教師データの浸入水率に対応する日時（時間帯）よりも「ａ分」だけ過去の時点での、当該着目している区域の降水量（ｍｍ）である。

The target rainfall (t) for the inflow treatment plant is the amount in a unit time for estimating the infiltration rate during rainy weather, for example, the accumulated value for one hour for an hourly estimate. However, in formula (5), k is the identifier of the area, and n is the total number of areas included in the target area. That is, in formula (5), the formula within the Σ symbol is the amount calculated for each area. Also, in formula (5), the "precipitation amount (mm) before the arrival time of the inflow" is the amount of precipitation (mm) in the area of interest at a time "a minutes" before the date and time (time period) corresponding to the infiltration rate of the teacher data to be calculated, for example, if the arrival time of the area of interest is "a minutes".

Note that the "flow arrival time" in formula (5) is the time it takes for infiltrated water, etc. to flow down the sewer from each area to reach its destination (sewage treatment plant), and is defined separately for each area, with one "flow arrival time" being defined for each area as the average value of the flow arrival times within one area. The flow arrival time for each area is determined in advance by calculation, measurement survey, etc. (in one example, the Manning formula is used to calculate the full pipe flow rate for each pipe from the pipe specifications in the sewerage register, the flow arrival time from the pipe to the treatment plant is determined, and the average value within each area is calculated and assigned to each area), and is stored in the memory unit 3 as part of the various data 19.

（６）式のとおり、区域の浸透率は、対象区域内のすべての面積について用途を決定し、用途ごとの区域の面積に占める割合と浸透率の積を集計したものである。

ただし、（８）式中、「区域の流入対象雨量」は上記（７）式で算出でき「処理場の流入対象雨量」は、上記（５）式で算出できる。また（８）式中、「雨天時流入量」とは、算出すべき教師データの浸入水率に対応する日時（時間帯）における、下水処理場の流入量の実測値であり、雨天時浸入水率推定装置１の要求に応じて下水処理場３２のコンピュータから雨天時浸入水率推定装置１へと送信される水位実測データ３９に含まれている（各種データ１９の一部として記憶部３に記憶される）。また（８）式中、「晴天時平均流入量」とは、季節、気象条件、曜日等に応じて推定可能な下水処理場の晴天時の平均流入量であり、事前に各種制御、表示部１３により推定値が算出されて各種データ１９の一部として記憶部３に記憶されているとする。変数データ取得部８は、このように教師データとしての浸入水率を算出して教師データ１７として記憶部３に格納する。

As shown in equation (6), the infiltration rate of an area is calculated by determining the uses for all areas within the target area, and then aggregating the product of the proportion of each use in the area and the infiltration rate.

However, in formula (8), the "inflow target rainfall of the area" can be calculated by the above formula (7), and the "inflow target rainfall of the treatment plant" can be calculated by the above formula (5). In formula (8), the "rainy weather inflow amount" is the actual measurement value of the inflow amount of the sewage treatment plant at the date and time (time period) corresponding to the infiltration rate of the teacher data to be calculated, and is included in the water level actual measurement data 39 transmitted from the computer of the sewage treatment plant 32 to the rainy weather infiltration rate estimation device 1 in response to a request from the rainy weather infiltration rate estimation device 1 (stored in the storage unit 3 as part of the various data 19). In formula (8), the "fine weather average inflow amount" is the average inflow amount of the sewage treatment plant on a fine day that can be estimated according to the season, weather conditions, day of the week, etc., and is assumed to be calculated in advance by the various control and display units 13 and stored in the storage unit 3 as part of the various data 19. The variable data acquisition unit 8 calculates the infiltration rate as teacher data in this way and stores it in the storage unit 3 as teacher data 17.

Examples of Machine Learning Algorithms Any known or unknown algorithm may be used as the machine learning algorithm, but here, a random forest and a neural network will be described as examples.

Figure 7 is a diagram explaining the concept of random forest (learning stage) as an example of a learning algorithm, and Figure 8 is a diagram explaining the concept of random forest (operation stage). An overview of random forest will be explained below. In the operation stage, the final model machine-learned by random forest uses multiple models called decision trees, and the final output is obtained by taking the majority vote (classification) and average (regression) of the prediction (estimation) results of each decision tree. In the learning stage of random forest, a large number of explanatory variables are classified into multiple subsamples by random restoration extraction using a method called bootstrap method, and a large amount of teacher data in each subsample is given to each decision tree, so that each decision tree learns separately and independently, and learning is performed with multiple models (decision trees). The final machine learning model generated by the machine learning algorithm of random forest can be interpreted as a collection of multiple decision trees.

Figure 9 is a diagram explaining the concept of a neural network as an example of a learning algorithm. The overview of a neural network will be explained below. In a neural network, one or more hidden layers (intermediate layers) exist between the input layer and the output layer, and the node values in the input layer are converted to node values in the hidden layer, and the node values in the hidden layer are converted to node values in the output layer, thereby obtaining output data (objective variable data) from input data (explanatory variable data). The conversion of node values from one layer to the next layer is performed by linear conversion or nonlinear conversion using an activation function. The connections between the nodes in the input layer and the hidden layer, and the connections between the nodes in the hidden layer and the output layer each have a separate weight value, and the weight values are updated by providing teacher data of the explanatory variables and the objective variables and learning. During learning, the weights of each layer are updated by the backpropagation method. The difference between the required output and the actual output is calculated to be small, and reflected in each layer. The model can be constructed arbitrarily by adjusting hyperparameters such as the number of intermediate layers and the number of nodes belonging to each intermediate layer. Random forests and neural networks are well-known machine learning algorithms and will not be described in further detail here.

In step S309 of executing machine learning , the machine learning unit 9 generates an estimation model of the rainwater infiltration rate by a machine learning algorithm using the teacher data 17 generated and stored in the previous steps. The machine learning unit 9 generates a trained model 16 by performing machine learning using the explanatory variable data in Table 1 and the objective variable data in Table 2 as teacher data, and stores the trained model 16 in the storage unit 3. When generating a model for each rainfall pattern, a model for rainfall pattern 1 is generated by performing machine learning using only the teacher data related to rainfall according to rainfall pattern 1 as teacher data, and similarly, a model for each rainfall pattern is generated by performing machine learning using only the teacher data related to rainfall according to the rainfall pattern as teacher data, and stores the model in the storage unit 3.

2. Operation during operation Figure 6 is a flowchart showing the operation flow of the operation stage executed by the rainfall infiltration rate estimation device. In the operation stage of the rainfall infiltration rate estimation device 1, when rainfall occurs and rainfall infiltration occurs in each area in the target area due to the rainfall, the infiltration rate in each area at each date and time is estimated using the trained model.

Setting Various Variables First, in step S601, the variable data acquisition unit 8 acquires data such as land use (permeability) as operation data of explanatory variables from the variable database 22. Specifically, the variable data acquisition unit 8 acquires variable data for each area from the variable database 22 as data on explanatory variables that can be regarded as constants that do not change at least on a short time scale other than the inflow target precipitation (or precipitation) among the explanatory variables in Table 1. Note that there are cases where variables are intentionally changed for the purpose of study, such as changing the pipe type or installation year.

In the rainfall data acquisition step S602, the variable data acquisition unit 8 acquires past rainfall data. Specifically, rainfall data is distributed in real time from the device of the precipitation measurement system 33. This distributed data is used to predict the inflow amount. The past data used to narrow down the rainwater infiltration occurrence area is obtained when an order is placed with the precipitation measurement system 33. After placing the order, the data is downloaded or received via an external storage medium. The variable data acquisition unit 8 acquires rainfall information data and predicted rainfall information data 40 from the computer of the precipitation measurement system 33 and stores them in the rainfall information database 20. In addition to actual rainfall data, virtual rainfall data may also be used.

Calculation of inflow target rainfall for each area Next, in step S603, the variable data acquisition unit 8 calculates the inflow target rainfall for each area. As in the case of creating the teacher data, the variable data acquisition unit 8 calculates the inflow target rainfall for each area according to the above formula (2). However, as the "precipitation amount (mm) of the area" in formula (2), the precipitation amount at the date and time to be estimated is obtained from the rainfall information database 20 and used instead of the precipitation amount in the teacher data.

Estimation of infiltration rate for each area By using the variable data value acquired in step S601 for each area and the value of the inflow target rainfall calculated in step S603 (if rainfall is used instead of the inflow target rainfall, the value of the rainfall acquired in step S602) as input data to the machine-learned model, an estimated value of the infiltration rate for each area can be obtained as a model output value. In step S604, the rainwater infiltration rate estimation unit 10 uses such input data to perform estimation using the above-mentioned machine-learning model to determine an estimated value of the infiltration rate for the target date and time in each area, and stores it in the rainwater infiltration rate database 23. In an example using random forest, if the final machine-learned model is composed of three decision trees, and the output (estimated value of infiltration rate) of each decision tree during operation is 5%, 10%, and 15%, the final output is a numerical value calculated from the three values and output from the random forest. For example, the average value of 10% is obtained as the estimated value of the infiltration rate. In one example using a neural network, the nine explanatory variables in Table 1 above belong to the input layer (for simplicity, only four nodes x1 to x4 are drawn in FIG. 9, but when nine explanatory variables are used, nine nodes x1 to x9 exist in the input layer and correspond to the respective explanatory variables), and the objective variable in Table 2 above belongs to the output layer (corresponding to node y1). In particular, when using an estimation model for each rainfall pattern, the rain-time infiltration rate estimation unit 10 determines which of rainfall patterns 1 to M the rainfall pattern of the rain to be predicted corresponds to from the rainfall information data acquired in step S602, and selects and uses the model for the rainfall pattern that is determined to correspond to the current rainfall pattern from the models for rainfall pattern 1 to rainfall pattern M stored as the trained models 16.

The estimated value of the infiltration water rate obtained by the output model estimation of the estimated value of the infiltration water rate is output in an arbitrary format by the various control and display units 13 in step S605. For example, on a map image divided into zones as shown in Fig. 5, a color (any expression, such as semi-transparent) corresponding to the estimated value of the infiltration water rate of each zone is assigned to each zone, and the map image is then displayed on the display device 27, or data of such a map image is transmitted to the client machine 34 and the map image is displayed on the display device of the client machine 34, so that a user such as an administrator can easily recognize zones with a high infiltration water rate.

Alternatively, instead of directly outputting the estimated value of the infiltration water rate, the estimated value of the infiltration water rate may be used by the rainwater infiltration water rate calculation unit 11 to calculate an estimate of the infiltration water rate during rain for each area, and the calculated estimated value of the infiltration water rate may be output in any format by the various control and display units 13 in step S605. Specifically, the rainwater infiltration water rate calculation unit 11 calculates an estimate of the infiltration water rate for the target date and time in each area by multiplying the inflow target rainfall in each area calculated in step S603 by the infiltration water rate estimated for each area in step S604 (see formula (1)). The various control and display units 13, for example, on a map image divided into zones as shown in Figure 5, assign a color (any expression, such as semi-transparent) to each zone corresponding to the estimated amount of water infiltration for that zone and then display the map image on the display device 27, or send data of such a map image to the client machine 34 and display the map image on the display device of the client machine 34, allowing users such as administrators to easily recognize areas with high amounts of water infiltration and narrow down the areas where water infiltration occurs during rainy weather.

3. Optimization of the machine learning model The machine learning unit 9 compares the estimated value of the infiltration rate with the actual measured value and updates the estimation model (in one example, machine learning is performed again using explanatory variable data during operation and the actual measured value of the infiltration rate as new teacher data (training data)). This can improve the performance of the estimation model. In particular, the machine learning unit 9 accumulates the variable data and the actual measured value of the infiltration rate (as an example of the actual measurement method, a flow meter is installed at the "midpoint" here for several months and the infiltration rate of the entire upstream area can be obtained by actual measurement) in the memory unit 3 as training data in the upstream area of the sewerage pipeline (in the example of FIG. 5, it is an area relatively far from the sewage treatment plant 48, for example, when each sewage pipe leading to the sewage treatment plant 48 is divided at the midpoint, the area through which the sewage pipe part farther from the sewage treatment plant 48 passes can be called the "upstream area"). The machine learning unit 9 provides the accumulated training data to the machine learning model and performs machine learning of the model again, which is considered to be an efficient optimization of the machine learning model.

により算出される区域の浸入水量を実測値として算出して用いてもよい。この場合、各区域の面積と対象地域の区域の面積の合計は、管理者等が予め測定しておくか、変数データ取得部８が地理情報データベース２４に格納された地理情報を用いて算出し、記憶部３の地理情報データベース２４に格納しておくとする。 As training data, the machine learning unit 9 may accumulate actual values of the amount of infiltration water together with the variable data. If the actual values of the amount of infiltration water are accumulated, the actual value of the infiltration water rate can also be calculated according to formula (1). The machine learning unit 9, the variable data acquisition unit 8, etc. may calculate the actual value of the amount of infiltration water from the above formulas (8) and (1), but it is also possible to calculate the actual value of the amount of infiltration water from the following formulas (9) and (10).

In this case, the total area of each zone and the area of the zone in the target area is measured in advance by the administrator or the like, or is calculated by the variable data acquisition unit 8 using the geographic information stored in the geographic information database 24, and is stored in the geographic information database 24 of the storage unit 3.

Here, in equation (9), the amount of infiltration water into the treatment plant may be calculated by the machine learning unit 9, the variable data acquisition unit 8, etc., using the following equation (11).

ただし、（１１）式中、ｋは区域の識別子であり、ｎは対象地域に含まれる区域の総数である。すなわち（１１）式において、Σ記号内の式は各区域について計算される量である。また（１１）式中、「流達時間前の降水量（ｍｍ）^※1」とは、例えば着目している区域の流達時間が「ａ分」であった場合、推定すべき浸入水率に対応する日時（時間帯）よりも「ａ分」だけ過去の時点での、当該着目している区域の降水量（ｍｍ）である（図１８中の、流達時間がそれぞれ０分、６２分、１２０分の例を参照）。

In equation (11), k is the identifier of the area, and n is the total number of areas included in the target region. That is, in equation (11), the formula within the Σ symbol is the amount calculated for each area. Also, in equation (11), "precipitation amount (mm) before the arrival time (mm) ^*1 " is the amount of precipitation (mm) in the area of interest at a time "a minutes" before the date and time (time period) corresponding to the infiltration water rate to be estimated, for example, if the arrival time of the area of interest is "a minutes" (see examples in Figure 18 where the arrival times are 0 minutes, 62 minutes, and 120 minutes, respectively).

Note that the "flow arrival time" in formula (11) is the time it takes for infiltration water during rain to reach the sewage treatment plant from each area, and is defined separately for each area, and one "flow arrival time" is defined for each area as the average value of the flow arrival times within each area. The flow arrival time for each area is determined in advance by measurement surveys, etc. (in one example, the full pipe flow rate is calculated for each pipe using the Manning formula from the pipe specifications in the sewerage register, the flow arrival time from the pipe to the treatment plant is obtained, and the average value within each area is calculated and assigned to each area), and is stored in the memory unit 3 as part of the various data 19.

4. Estimation of infiltration water inflow volume at sewage-related facilities Next, a method for predicting in real time the infiltration water inflow volume into sewage-related facilities such as the sewage treatment plant 48 using the infiltration water rate estimated by the above-mentioned estimation model will be described. As a premise, as already mentioned, the rainfall information database 20 stores precipitation (mm per hour) data for each area at each time in association with date and time, and in addition, it is assumed that the predicted value of future precipitation in the rain is also stored in the rainfall information database 20 as predicted precipitation (mm) for each area at each time in association with date and time (it is assumed that the variable data acquisition unit 8 receives predicted precipitation data from the computer of the precipitation measurement system 33 and stores it in the rainfall information database 20). Also, during a certain rainfall, the infiltration rate in each area is estimated from time to time as described above using Figures 3 to 6, and the estimated value of the infiltration rate for each area at each time is stored in the rainwater infiltration rate database 23 in association with the date and time. Furthermore, in an area where the runoff time is shorter than the difference between the calculation start time and the time to be predicted (for example, a 60-minute prediction), the measured rainfall value cannot be used for prediction, so the rainfall data for the selected area is used. Using the actual rainfall value and the predicted future rainfall value stored in the rainfall information database 20, the infiltration rate for each area at each future time is estimated using a machine-learned estimation model, and the estimated value of the infiltration rate for each area at each future time is stored in the rainwater infiltration rate database 23 in association with the date and time. In addition, it is assumed that the estimated sunny weather inflow for each area at each time has already been stored in memory unit 3 in association with the date and time as part of the various data 19. However, since the sunny weather inflow for each area is an amount that can be estimated based on the season, weather conditions, day of the week, etc. and is treated as a known amount (an estimated value has already been obtained), it is assumed that the estimated sunny weather inflow for each area at each future time ("sunny weather inflow" is the inflow amount when there is no water infiltration during rain (inflow that is not caused by rain can occur on sunny days, but this is included as it is based on the treatment plant inflow amount). Here, the target is indicated by "for each area") has already been stored in memory unit 3 as part of the various data 19.

Figure 10 is a diagram explaining the concept of estimating the amount of inflow to a sewage-related facility at a future time after a reference time. The amount of inflow (in the present example, the inflow amount per hour (t) is the same as the inflow target rainfall and infiltration water amount) flowing into a sewage-related facility such as a sewage treatment plant 48 depends on the amount of rainfall in each area multiplied by the infiltration water rate, but since it takes different amounts of time for the inflow water from each area to reach the sewage-related facility, what contributes to the amount of inflow to a sewage-related facility at a specific time is not the specific time, but the amount of precipitation in each area at a time before the specific time by the amount of inflow time that differs for each area (the amount of precipitation at different times for each area needs to be taken into account). Note that "time" here refers to a "time" that also specifies the "day", that is, "date and time".

ただし、（１２）式中、ｋは区域の識別子であり、ｎは対象地域に含まれる区域の総数である。すなわち（１２）式において、Σ記号内の式は各区域について計算される量である。また（Ｔ（分）－流達時間（分））がマイナスの値となる場合、（１２）式中の「予測降水量（ｍｍ）」としては実績降雨による降水量（ｍｍ）を用いる。また（１２）式中の浸入水率は推定モデルによる推定値であるが、推定モデルを降雨パターン別に生成する場合には、流入量の推定の対象とする降雨の降雨パターンに対応するパターン別モデルを用いて浸入水率の推定値を決定すればよい。 Specifically, the hourly inflow amount (t: tons) of infiltration water flowing into a sewage-related facility such as a sewage treatment plant 48 T minutes after a certain reference point (T=0) is estimated by the following equation (12).

In equation (12), k is the identifier of the area, and n is the total number of areas included in the target region. That is, in equation (12), the formula within the Σ symbol is the amount calculated for each area. Furthermore, when (T (min) - arrival time (min)) is a negative value, the precipitation amount (mm) from actual rainfall is used as the "forecasted precipitation amount (mm)" in equation (12). Furthermore, the infiltration water rate in equation (12) is an estimate based on an estimation model, but when an estimation model is generated by rainfall pattern, the estimated infiltration water rate can be determined using a pattern-specific model corresponding to the rainfall pattern of the rainfall to be estimated for the inflow amount.

The sewage-related facility inflow amount estimation unit 12 reads out the necessary data from the rainfall information database 20, the rainwater infiltration rate database 23, the various data 19, etc., and estimates the inflow amount to the sewage-related facility according to the formula (13). The estimated inflow amount may be displayed on the display device 27 by the various control and display units 13, or the estimated inflow amount and an identifier for identifying the sewage-related facility may be sent to the client machine 34 or the computer of the sewage-related facility 32 to be displayed on the input/output unit (display unit) 43 of the client machine 34 or the display device of the computer of the sewage-related facility 32. In this way, if the inflow amount of inflow water during rainy weather to the sewage-related facility is predicted, it can contribute to the stable and efficient operation of pumping stations, sewage treatment plants, etc. In principle, the inflow amount of the separate sewer system is not rainwater, and even in rainy weather, it is planned to be the same as in sunny weather. However, depending on the municipality, a large amount of rainwater may flow in during rainy weather, causing problems in the operation of the pumping station or sewage treatment plant. Although stormwater drainage facilities and combined sewers have the capacity to remove stormwater, when the inflow exceeds the capacity or when there is a large fluctuation in the inflow, it places a heavy burden on the operation of the facilities. By applying the system for narrowing down the area where inflow water occurs during rain, it is possible to predict the inflow amount flowing into a pumping station or sewage treatment plant in real time during rain, making it possible to catch large inflows in a timely manner. Another advantage of inflow prediction is that, in principle, the inflow gate is closed to deal with an inflow amount that cannot be treated, but if the timing is delayed, the facility will be flooded and the treatment function will be lost, leading to flooding of the surrounding area, as well as high costs and long-term restrictions on sewage treatment. However, even if the inflow gate is closed, the surrounding area will be flooded, so the gate cannot be closed preventively. As mentioned above, by accurately predicting the inflow, it is possible to minimize flooding of the surrounding area while ensuring the safety of the treatment plant. By predicting the inflow amount in the near future, it is possible to appropriately control the operation of the pump when the inflow is on a decreasing trend. This reduces the workload of the operators and energy consumption.

11 to 13 show screen images of the system for narrowing down the occurrence area of water infiltration during rainy weather. These screens may be displayed on the display device 27 of the water infiltration rate estimation device during rainy weather 1, or the screen display data may be transmitted from the water infiltration rate estimation device during rainy weather 1 to the client machine 34 and then displayed on the display device of the input/output unit (display unit) 43 of the client machine 34.

Performance evaluation results of the rainwater infiltration rate estimation model created as an embodiment of the present invention are described below. An estimation model (analysis model) was constructed using rainfall information from January to November 2019 and treatment plant inflow information. The explanatory variables (input items for each area) of this estimation model are as shown in Table 3 below. However, each area was assumed to be a square with one side of 250 mm, and the model was constructed using rainfall information from XRAIN radar. The objective variable was the infiltration rate.

（表３）

(Table 3)

The analysis procedure is as follows.
(1) Estimation of inflow volume on sunny days The inflow volume on sunny days for the entire target area is calculated from the inflow volume to the treatment plant on sunny days, and the inflow volume on sunny days is allocated to each mesh. At that time, the building area within the mesh is obtained by image analysis of the map, and used to set the wastewater volume basic unit.
(2) Input of rainfall information: XRAIN rainfall information is assigned to each area (mesh).
(3) Input of watershed characteristics (variables) Various watershed characteristic values such as the infiltration condition at the ground surface, the length of sewage pipes and storm drains, etc. are assigned to each mesh.
(4) Input of flow arrival time The full pipe flow velocity is calculated for each pipe using the Manning's formula from the pipe specifications in the sewerage register, and the flow arrival time from the pipe to the treatment plant is obtained. The average value within the mesh is calculated and assigned to each mesh.
(5) Input of inflow volume to treatment plant Import the inflow volume data to the treatment plant (hourly or more detailed data).
(6) Analysis of infiltration volume and infiltration rate during rainy weather Wastewater and infiltration water during rainy weather from each mesh flows into the treatment plant after the flow time of that mesh has elapsed. The infiltration rate of each mesh can be calculated by analyzing the relationship between the fluctuation in rainfall in each mesh and the fluctuation in inflow to the treatment plant using machine learning.

The examination flow for this performance evaluation is as follows:
1. Organizing existing information - Rainfall (XRAIN)
・Geographical characteristics, facility characteristics, etc. ・Treatment plant inflow volume 2. Model construction ・Learn using existing information 3. Analysis of infiltration rate ・Input rainfall data from the flow survey period into the constructed model and calculate the infiltration rate 4. Verification ・Compare with actual measurements (flow survey meter at 10 locations)

To build the analytical model, rainfall information from January to November 2019 and treatment plant inflow information were used, and the verification rainfall for comparison with the actual measured values was the rainfall on October 19, 2019 during the flow survey period (total rainfall 75 mm, maximum hourly rainfall 12 mm/h). The analysis results of the maximum infiltration water rate during the verification rainfall are shown in Figure 14.

（表４） In addition, if the items of the watershed characteristics are arranged in order of their degree of relevance to the infiltration water rate, the results are as shown in Table 4 below.

(Table 4)

As shown in Table 4, the watershed characteristic with the highest correlation with infiltration water rate (importance of features in random forest) is "rainfall", with a correlation of 0.453641. The watershed characteristic with the second highest correlation is "residential area", with a correlation of 0.166464. The watershed characteristic with the third highest correlation is "infiltration rate", with a correlation of 0.081008.

Figures 15 and 16 show the results of comparing the analysis results of the infiltration rate with the actual values calculated from the flow rate survey results. The flow rate survey was conducted for approximately two months from September 2019, with flow meters installed in a total of 10 locations (A to J) in the pipe. In the graph in Figure 15, for each of A to J, the left bar graph is the actual measurement value and the right bar graph is the analysis result (estimated value using the estimation model). The scale of the vertical axis of the bar graph corresponds to the "infiltration rate (%)" on the left, and the scale on the right side of the vertical axis corresponds to the total rainfall shown between the actual measurement value and the analysis result bar graph. The comparison was performed using the average infiltration rate for three hours from 4:00 to 7:00, when the rainfall was heavy in the verification rainfall. As a result of the comparison, the infiltration rate of the analysis result was within ±10% of the actual measurement value in six locations and within ±20% in two locations, which was roughly the same as the actual measurement value.

The present invention can be used to narrow down the areas where water infiltrates during rain in separate sewer systems, to improve the efficiency of maintenance such as repairing and replacing sewer pipes (sewage pipes), and to predict the amount of inflow during rain into sewerage-related facilities including combined sewer systems and storm water drainage facilities, but it can also be used in a wide range of applications without being limited to these.

1 Rainfall water infiltration rate estimation device 2 Control unit 3 Memory unit 4 Input/output unit 5 Communication unit 6 Processor 7 Temporary memory 8 Variable data acquisition unit 9 Machine learning unit 10 Rainfall water infiltration rate estimation unit 11 Rainfall water infiltration amount calculation unit 12 Sewerage-related facility inflow amount estimation unit 13 Various control and display programs 14 Machine learning related programs 15 Various programs 16 Trained model 17 Teacher data 18 Test data 19 Various data 20 Rainfall information database 21 Facility inflow amount, rainfall water infiltration amount database 22 Variable database 23 Rainfall water infiltration rate database 24 Geographic information database 25 Keyboard 26 Mouse 27 Display device 28 Communication interface 29 Communication circuit 30 Communication line (Internet, network line, etc.)
31 External server machine (geographic information server)
32 Sewage-related facilities (sewage treatment plants, pumping stations, etc.)
33 Precipitation measurement systems (weather radar, rain gauges, etc.)
34 Client machines (personal computers, smartphones, etc.)
35 Map data 36 Land use data 37 Ground rain gauge position data 38 Other geographical data 39 Water level measurement data 40 Rainfall information data 41 Control unit 42 Memory unit 43 Input/output unit 44 Communication unit 45 Mesh structure 46 Mesh (area)
47 mesh (area)
48 Sewage (sewage) treatment plant 49 Mesh (area)
50 Sewage pipe (inlet pipe)
51 mesh (area)
52 Sewage pipe (inlet pipe)
1000 Sewer pipes (sewage pipes)
1001 Sewage (sewage) treatment plant 1002 Defective parts (cracks, breaks, holes, etc.)
1100 Storm water pipe 1101 Storm water pumping station 1200 Building (private house)
1300 Buildings (Office Buildings)
1400 Rain clouds

Claims (7)

A rainwater infiltration rate estimation device that estimates a rainwater infiltration rate that corresponds to a ratio of rainwater infiltration volume to rainfall, the ratio being defined for each of a plurality of zones in a target area, comprising:
A variable data acquisition unit that acquires data of a variable defined for each of the areas, the variable including a function of precipitation in each of the areas;
A rainwater infiltration rate estimation unit,
data of the variables in each of the zones acquired by the variable data acquisition unit;
a rainwater infiltration rate estimation unit that estimates the rainwater infiltration rate in each of the areas using an estimation model that estimates the rainwater infiltration rate from the variables, the estimation model being obtained by performing machine learning using past data on the rainwater infiltration rate and the variables as teacher data ;
The estimation model is a pattern-specific estimation model corresponding to the rainfall pattern of the target rainfall, among two or more separate pattern-specific estimation models each of which is separately machine-learned corresponding to two or more rainfall patterns using past data of the rainfall infiltration rate and the variables as teacher data,
The rainfall pattern is a pattern of locally concentrated rainfall, and includes a front concentration pattern and a rear concentration pattern . The rainwater infiltration rate estimation device according to claim 1, wherein the machine learning learning algorithm is a random forest or a neural network. The rainwater infiltration rate estimation device of claim 1 or 2, further comprising a rainwater infiltration volume calculation unit that calculates an estimated value of the rainwater infiltration volume in each of the areas using data on inflow target rainfall, which is a function of the precipitation in each of the areas acquired by the variable data acquisition unit, and an estimated value of the rainwater infiltration rate in each of the areas obtained by estimation by the rainwater infiltration rate estimation unit. The rainwater infiltration rate estimation device according to claim 1 , wherein the function of precipitation is an inflow target rainfall calculated using the precipitation and an area of each of the regions. The rainwater infiltration rate estimation device according to any one of claims 1 to 4 ,
a rainwater infiltration rate storage unit that stores the estimated rainwater infiltration rate in each of the areas obtained by the rainwater infiltration rate estimation unit in association with date and time;
A sewage-related facility inflow amount estimation unit that estimates an inflow amount of inflow water flowing into a sewage-related facility during rainy weather at a future time point after a certain time from a reference time point,
A predicted precipitation amount or actual precipitation amount for each area at a date and time determined for each area according to the flow arrival time from each area to the sewerage-related facility;
the area of each of said regions; and
a permeability rate for each of said regions; and
An estimate of the rainwater infiltration rate of each of the areas at the date and time determined for each of the areas according to the runoff time;
and a sewage-related facility inflow estimation unit that estimates an inflow at a future point in time that is a certain time period after the reference point in time using the clear weather inflow at the date and time estimated for each of the areas according to the inflow arrival time. A method for estimating a rainwater infiltration rate, which is defined for each of a plurality of areas in a target area divided into the plurality of areas and corresponds to a ratio of an amount of rainwater infiltration to an amount of rainfall, is performed by a rainwater infiltration rate estimation device,
a variable data acquisition step of acquiring data for variables defined for each of the zones, the variables including a function of precipitation in each of the zones;
A rainwater infiltration rate estimation process,
data of the variables in each of the areas acquired in the variable data acquisition step;
and a rainwater infiltration rate estimation process for estimating the rainwater infiltration rate in each of the areas using an estimation model that estimates the rainwater infiltration rate from the variables, the estimation model being obtained by performing machine learning using past data on the rainwater infiltration rate and the variables as teacher data ;
The estimation model is a pattern-specific estimation model corresponding to the rainfall pattern of the target rainfall, among two or more separate pattern-specific estimation models each of which is separately machine-learned corresponding to two or more rainfall patterns using past data of the rainfall infiltration rate and the variables as teacher data,
The rainfall pattern is a pattern of locally concentrated rainfall, and includes a front concentration pattern and a rear concentration pattern . A program for causing a computer to execute the method according to claim 6 .

Images