KR102369350B1 - Predicting System of Energy Consumption and Data of Air Conditioning Equipment Using Artificial Neural Network and Method Thereof - Google Patents

Abstract

Disclosed are a predicting system of energy consumption and data of an air conditioning equipment using an artificial intelligence (AI) neural network and a method thereof. According to an embodiment of the present invention, the method comprises: a history data collection step to collect history data from one or more of a BEMS, BAS, and air conditioning equipment; a system input step to receive a lower system of the air conditioning equipment desired to be predicted from a user; a list provision step to provide an input variable list on the lower system to the user; a variable input step to receive the input of the input variable selected by the user among the input variable list; a neural network model drawing step to automatically normalize the history data, automatically optimize the neural network model, and draw a final neural network model based on the input variable and a preset value in a certain neural network model; and a prediction value drawing step to draw one or more of energy consumption prediction value and data prediction value of the lower system by using the final neural network model. The present invention aims to provide a predicting system of energy consumption and data of an air conditioning equipment using an AI neural network and a method thereof, which are capable of allowing even a user with no special knowledge to easily predict energy consumption and related data of the air conditioning equipment.

Description

Predicting System of Energy Consumption and Data of Air Conditioning Equipment Using Artificial Neural Network and Method Thereof

The present disclosure relates to a system and method for predicting energy usage and data of an air conditioning facility using an artificial neural network.

The content described in this section merely provides background information for the present disclosure and does not constitute the prior art.

Currently, as building intelligence and informationization progresses, building management systems such as IBS and BAS BEMS are being installed in buildings, and accordingly, the amount of information that can be obtained from buildings is increasing. In particular, recently, the installation of BEMS in public buildings has become compulsory, and accordingly, the market related to the building management system is expected to gradually expand to buildings in the private sector.

On the other hand, analyzing and predicting the amount of energy used in a building is very important for saving and managing energy in a building. In particular, in the regulation for confirming the installation of BEMS, it is mandatory to perform the function of predicting the energy consumption of a building.

In the building energy management system (BEMS) installation regulations, the method of estimating energy consumption is sequentially classified into a method of predicting through a simple comparison of the previous year or the previous month, a method using statistical techniques such as regression analysis, and other advanced methods. there is.

In addition, depending on the energy consumption estimation method being performed in the building, the points are allocated when the BEMS is installed in the building and then installed when the installation is confirmed. For example, a method of predicting through simple comparison has a low score compared to a method using a statistical technique, and a method using a statistical technique has a low score compared to other advanced methods.

On the other hand, the air conditioning equipment in charge of heating and cooling has a characteristic that the operating time, operating conditions, etc. change very frequently depending on the climate. Accordingly, compared to other facilities such as lighting facilities, water supply and drainage facilities, transportation facilities, etc., energy consumption for air conditioning facilities may cause many more errors in simple comparison with the previous year and the previous month or when predicting with statistical analysis methods. can Therefore, recently, precise methods using artificial intelligence techniques are being used. For example, a prediction method using an artificial neural network is representative.

In order to use an artificial neural network, there are three main steps. The first is the process of selecting input variables necessary for neural network learning, the second is the process of normalizing the data collected as input values to facilitate the learning process using the artificial neural network, and the third is the process of optimizing the neural network structure.

For example, adjusting the number of neurons and the number of hidden layers of the initially created neural network model, or repeatedly modifying the optimization process of input variables and structural variables can be regarded as such processes.

After that, the process of normalization of input data and model optimization is checked from the initial neural network model created, and after evaluation through prediction performance, it is used for actual energy prediction.

This series of processes requires a lot of experience and understanding of the building system along with expertise in the field of artificial intelligence, and it requires a lot of effort to apply it to an actual system.

Accordingly, the present disclosure provides a system and method for predicting energy use and data of an air conditioning system using an artificial neural network, which enables users without professional knowledge to easily predict energy consumption and related data of the air conditioning system. has the main purpose of

According to an embodiment of the present disclosure, a history data collection process of collecting history data from at least one of BEMS, BAS, and air conditioning equipment; a system input process of receiving a sub-system of an air conditioning system that a user wants to predict; a list providing process of providing a user with a list of input variables related to the subsystem; a variable input process of receiving an input variable selected by a user from a list of input variables; A neural network model derivation process of deriving a final neural network model by automatically normalizing historical data and automatically optimizing the structure of the neural network model based on input variables and preset values for a predetermined neural network model; and a predicted value derivation process of deriving at least one of an energy usage predicted value and a data predicted value of a subsystem by using the final neural network model.

As described above, according to the present embodiment, the method of predicting the energy consumption and data of the air conditioning equipment has an effect of enabling a user without professional knowledge to easily predict the energy consumption and related data of the air conditioning equipment.

1 is a flowchart of a method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure.
2 is an exemplary diagram of a list of input variables provided to a user in a method for predicting energy usage and data of an air conditioner according to an embodiment of the present disclosure.
3 is an exemplary view illustrating a state in which an energy usage predicted value is output in the method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure.
4 is a block diagram of a system for predicting energy usage and data of an air conditioner according to an embodiment of the present disclosure.

Hereinafter, some embodiments of the present disclosure will be described in detail with reference to exemplary drawings. In adding reference numerals to components in each drawing, it should be noted that the same components are given the same reference numerals as much as possible even though they are indicated in different drawings. In addition, in describing the present disclosure, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present disclosure, the detailed description thereof will be omitted.

In describing the components of the embodiment according to the present disclosure, reference numerals such as first, second, i), ii), a), b) may be used. These signs are only for distinguishing the elements from other elements, and the essence, order, or order of the elements are not limited by the signs. In the specification, when a part 'includes' or 'includes' a certain element, it means that other elements may be further included, rather than excluding other elements, unless explicitly stated otherwise. .

1 is a flowchart of a method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure.

2 is an exemplary diagram of a list of input variables provided to a user in a method for predicting energy usage and data of an air conditioner according to an embodiment of the present disclosure.

3 is an exemplary diagram illustrating a state in which an energy usage predicted value is output in the method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure.

Referring to FIG. 1 , the method of predicting energy consumption and data of an air conditioning facility is a history data collection process (S101), data transmission process (S102), system input process (S103), list provision process (S104), variable input process ( S105), a model variable input process (S106), a neural network model derivation process (S107), a predicted value derivation process (S108), and a predicted value output process (S109) may be included.

History data for energy usage and data prediction of air conditioning equipment may be collected from at least one of BEMS, BAS, and air conditioning equipment (S101).

The history data is data related to the air conditioning equipment accumulated for a certain period in the past, and may include actual measured values of the air conditioning equipment, energy consumption of the air conditioning equipment, indoor temperature, indoor humidity, and the like.

For example, the history data for the refrigerator includes data such as the outdoor wet bulb temperature, the outdoor air dew point temperature, the cold water/supply temperature, the cold water/hot water flow rate, the cooling water feed water temperature, etc. measured during a specific period in the past, and the energy consumption of the refrigerator. It may contain data.

The historical data may be collected from a database stored in a building management system such as BEMS or BAS, or may be directly collected from each sub-system of the air conditioning system.

Meanwhile, in this specification, the data predicted value refers to a predicted value for a series of data related to air conditioning of a building. For example, the data predicted value may include predicted values for indoor temperature, indoor humidity, cold water supply temperature, air volume, boiler hot water temperature, and the like.

Accordingly, the method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure is effective in predicting not only the energy usage of the air conditioning facility, but also other data required for air conditioning of a building or management of a building there is

Referring back to FIG. 1 , after the process S101 , the history data collected in the history data collection process S101 may be transmitted to the server ( S102 ). The server may be a web server or a local server.

In the case of the embodiment in which the history data is stored in the web server, input and output of data may be made in a web format. For example, a user may input a variable for energy usage prediction in a web format through an application installed on a user terminal (eg, a smartphone), and an energy usage prediction value derived through a neural network model will also be provided through the application. can

Thereafter, the user may directly input or select a subsystem of the air conditioning system to be predicted through the neural network model (S103).

Here, the sub-system refers to each system constituting the air conditioning system. For example, the subsystem may be a heat source facility such as a refrigerator, a boiler, a geothermal heat pump, an ice heat storage system, or may include an air conditioner (AHU).

Thereafter, the user may be provided with a list of input variables related to the subsystem selected in the system input process (S103) (S104).

Referring to FIG. 2 , the list of input variables provided to the user may include a series of variables necessary for predicting energy usage.

For example, if the subsystem selected by the user is a heat source facility such as a refrigerator, a boiler, or a geothermal heat pump, the input variable list provided to the user includes the date, time, outside air wet bulb temperature, outside air dew point temperature, cold water supply temperature, And, at least one of cold water supply flow rate may be included.

If the sub-system selected by the user is the air conditioner, the list of input variables provided to the user may include at least one of date, time, outdoor air wet bulb temperature, outdoor air dew point temperature, air supply air volume, and supply air temperature.

The above-described variables are for the convenience of description and only a part of the variables are described. Accordingly, in the present disclosure, the input variable list provided to the user may include other variables in addition to the aforementioned variables. Of course, these variables may vary depending on the subsystem selected by the user.

The user may directly input or select some of the plurality of variables included in the input variable list (S105). The input variable selected by the user may be set as an input variable of the neural network model.

In addition, the user may input at least one of a variable required when designing a neural network model, for example, an interval of input data, the number of neurons, and an artificial neural network learning period ( S106 ).

Referring to FIG. 2 , variables necessary for designing a neural network model may also be included in the input variable list and provided to the user. The user can input input variables related to the subsystem and variables necessary for designing the neural network model from the provided input variable list.

In the method of predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure, when a user selects a sub-system of an air-conditioning facility, by providing major variables for the sub-system in a list format, a user without specialized knowledge It also has the effect of being able to easily select input variables for neural network model design.

On the other hand, in the case of the embodiment in which the history data is stored in the web server, the above-described processes ( S103 to S106 ) may be performed in a web format.

For example, the provision of the input variable list or the user's data input may be performed in a web format through the user's PC or a user terminal (eg, a smart phone).

Meanwhile, in FIG. 1 , process S106 is illustrated as being performed after process S105, but this is for convenience of description, and the order of the present disclosure is not limited thereto. Accordingly, step S106 may be performed before step S105 or may be performed simultaneously with step S105.

Referring back to FIG. 1, after the process S105 or S106, based on the input variable input from the user and the preset value for the predetermined neural network model, the historical data can be automatically normalized, and the structure of the neural network model is can be automatically optimized. Accordingly, the final neural network model can be derived (S107).

In step S107 , the neural network model or preset values used for normalization of historical data and optimization of the neural network model may use an existing verified neural network model and default values of the neural network model.

Therefore, the method for predicting energy usage and data of an air conditioning facility according to an embodiment of the present disclosure is to automatically perform normalization of historical data and optimization of the neural network model using the existing verified neural network model and default values of the neural network model. By doing so, there is an effect that a user without specialized knowledge can easily design a neural network model.

Meanwhile, in the neural network model derivation process ( S107 ), the user may be partially involved in the normalization of historical data or optimization of the neural network model by adding or deleting related variables in order to increase the predictive performance of the neural network model.

Meanwhile, the final neural network model derived through the process S107 may include a nonlinear autoregressive exogenous (NARX) model.

The NARX model consists of an input layer, a hidden layer, and an output layer. In the input layer, the input value neural network learning for each subsystem is performed. The input signal of the input layer is transmitted to the hidden layer and neural network operation is performed by the internal neurons. The operation result of the hidden layer is output through the output layer. In order to improve the energy consumption prediction accuracy, optimization of key learning parameters such as a hidden layer, a neuron, an epoch, and a learning rate may be performed.

The NARX model is known to have high predictive performance for time-series data. Since the energy consumption of air conditioning equipment is time-series data, the NARX model can be suitable as a neural network model for predicting energy consumption. However, the present disclosure is not limited thereto, and the final neural network model according to the present disclosure may include a neural network model other than the NARX model.

Thereafter, at least one of the energy usage prediction value and the data prediction value of the subsystem may be derived or calculated using the final neural network model ( S108 ).

The energy usage predicted value may be derived for a specific period or a specific time. For example, the energy usage predicted value may be derived in time series in the manner of 'energy usage for a month in September 2020'.

Thereafter, at least one of the energy usage predicted value and the data predicted value may be output (S109). In this case, the output energy usage prediction value or data prediction value may be expressed as a monthly time series.

Referring to FIG. 3 , the energy consumption predicted value may be output together with the energy consumption of the previous year corresponding to the same period of the previous year or the same time of the previous year. For example, the predicted energy usage value for the month of September 2020 may be output together with the energy usage (actual value) for the month of September 2019.

In addition, the energy usage predicted value may be output together with the previous year's energy usage predicted value derived through the final neural network model for the same period of the previous year or the same time of the previous year. For example, the energy usage (measured value) for the month of September 2019 may be output together with the energy usage predicted value for the month of September 2019 derived through the final neural network model.

The user may determine the prediction performance of the final neural network model by comparing the predicted energy usage value with the previous year's energy usage or comparing the previous year's energy usage with the previous year's energy usage predicted value. If the determined prediction performance of the final neural network model does not meet expectations, the user may improve the prediction performance of the final neural network model by repeatedly performing all or part of steps S103 to S109.

Meanwhile, the prediction value output process ( S109 ) may be performed in a web format. For example, at least one of the energy usage predicted value and the data predicted value may be output through the user's PC or a user terminal (eg, a smart phone).

4 is a block diagram of an energy usage and data prediction system 40 of an air conditioning facility according to an embodiment of the present disclosure.

Referring to FIG. 4 , the energy usage and data prediction system 40 of the air conditioning facility includes a history data collection unit 410 , a data input unit 420 , a list providing unit 430 , a neural network model derivation unit 440 , It may include a prediction value derivation unit 450 and a prediction value output unit 460 .

The history data collection unit 410 may be configured to collect history data from at least one of BEMS, BAS, and air conditioning equipment. The history data collected by the history data collection unit 410 may be transmitted to a web server or a local server.

The data input unit 420 may be configured to receive input variables of the sub-system of the air conditioning equipment desired by the user and the sub-system selected by the user.

The list providing unit 430 may be configured to provide a list of input variables related to the subsystem to the user.

The neural network model derivation unit 440 may be configured to derive the final neural network model by automatically normalizing historical data and automatically optimizing the structure of the neural network model based on input variables and preset values for a predetermined neural network model. there is.

The predicted value derivation unit 450 may be configured to derive an energy usage predicted value and a data predicted value of a subsystem by using the final neural network model.

The predicted value output unit 460 may be configured to output at least one of an energy usage predicted value and a data predicted value. The predicted value output unit 460 may be the user's PC or the user terminal T.

In FIG. 4, each component (410 to 460) of the energy usage and data prediction system 40 of the air conditioning facility is shown separately, but this is for convenience of explanation, and each component is implemented as a separate and separate member doesn't mean it will be Accordingly, some of the respective components 410 to 460 may be implemented together within one module or one physical structure.

The above description is merely illustrative of the technical idea of this embodiment, and a person skilled in the art to which this embodiment belongs may make various modifications and variations without departing from the essential characteristics of the present embodiment. Accordingly, the present embodiments are intended to explain rather than limit the technical spirit of the present embodiment, and the scope of the technical spirit of the present embodiment is not limited by these embodiments. The protection scope of this embodiment should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be interpreted as being included in the scope of the present embodiment.

410: history data collection unit 420: data input unit
430: list providing unit 440: neural network model derivation unit
450: predicted value derivation unit 460: predicted value output unit

Claims (17)

a history data collection process of collecting historical data from at least one of BEMS, BAS, and air conditioning equipment;
a system input process of receiving an input from a user of a subsystem of an air conditioning system desired for prediction;
a list providing process of providing a list of input variables related to the subsystem to the user;
a variable input process of receiving an input variable selected by the user from the input variable list;
a neural network model derivation process of deriving a final neural network model by automatically normalizing the historical data and automatically optimizing the structure of the neural network model based on the input variables and preset values for a predetermined neural network model; and
A prediction value derivation process of deriving at least one of an energy usage prediction value and a data prediction value of the subsystem using the final neural network model,
In the system input process and the variable input process,
The sub-system and the input variable are energy usage and data prediction method of an air conditioning facility, characterized in that input through a user terminal. According to claim 1,
Between the historical data collection process and the system input process,
The method of predicting energy usage and data of an air conditioning facility, characterized in that it further comprises a data transmission process of transmitting the history data to the server. delete According to claim 1,
After the prediction value derivation process,
The method of predicting energy consumption and data of an air conditioning facility further comprising a predicted value output process of outputting at least one of the predicted energy usage value and the data predicted value. 5. The method of claim 4,
In the process of deriving the predicted value,
The energy usage prediction value is a method of predicting energy usage and data of an air conditioning facility, characterized in that it is derived for a predetermined period or a predetermined time. 6. The method of claim 5,
In the process of outputting the predicted value,
The energy usage prediction value is the energy usage and data prediction method of the air conditioning facility, characterized in that the output together with the energy usage of the previous year corresponding to the same period of the previous year or the same time of the previous year. 6. The method of claim 5,
In the process of outputting the predicted value,
The energy usage prediction value is, for the same period of the previous year or the same time of the previous year, the energy usage and data prediction method of an air conditioning facility, characterized in that it is output together with the energy usage prediction value of the previous year derived through the final neural network model. 5. The method of claim 4,
In the process of outputting the predicted value,
At least one of the predicted energy usage value and the predicted data value is an energy usage and data prediction method of an air conditioning facility, characterized in that it is output through a user terminal. According to claim 1,
Before the neural network model derivation process,
The method of predicting energy consumption and data of an air conditioning system, further comprising a model variable input process of receiving at least one of an interval of input data, the number of neurons, and an artificial neural network learning period from the user. According to claim 1,
The final neural network model is a method of predicting energy use and data of an air conditioning system, characterized in that it includes a nonlinear autoregressive exogenous (NARX) model. According to claim 1,
The sub-system of the air conditioning system, energy usage and data prediction method of the air conditioning system, characterized in that it includes at least one of an air conditioner, a refrigerator, a boiler, and a geothermal heat pump. 12. The method of claim 11,
The input variable for the air conditioner is,
A method for predicting energy consumption and data of an air conditioning system, comprising at least one of date, time, outdoor air wet bulb temperature, outdoor air dew point temperature, supply air volume, and supply air temperature. 12. The method of claim 11,
The input variable for at least one of the refrigerator, the boiler, and the geothermal heat pump,
Date, time, outdoor air wet bulb temperature, outdoor air dew point temperature, cold water supply temperature, and the energy consumption and data prediction method of an air conditioning facility, characterized in that it comprises at least one of the flow rate of cold water. According to claim 1,
The data prediction value is an energy usage and data prediction method of an air conditioning facility, characterized in that it includes a predicted value for at least one of indoor temperature and indoor humidity. delete delete delete

Images