JP3932383B2 - Traveling time prediction device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a road traffic system and relates to a travel time prediction device for predicting the travel time of an arbitrary section of a target road from a departure point to an arrival point.
[0002]
[Prior art]
Conventionally, on a highway or the like, the time required to travel from one point to another point, that is, information on the required travel time is spatial average speed information obtained from a sensor such as a vehicle detector installed on the road. It was demanded from.
[0003]
When a vehicle detector is used, a target section with a target route is divided into a plurality of unit sections, a vehicle detector is installed in each unit section, and each unit section is measured using the traveling vehicle speed measured by the vehicle detector. Calculate travel time. The required travel time information of the target section of the target route is obtained by summing the calculated travel time of each unit section.
[0004]
In addition, an AVI system that measures the actual travel time of each vehicle that actually traveled in the target section by recognizing the number of the traveling vehicle at both ends of the target section has been put into practical use.
[0005]
However, such a method using a sensor such as a vehicle detector or a vehicle number image recognition device requires that the sensor is installed, and the travel time information is not provided on a route where the sensor is not installed. I can't get it. Also, depending on sensors such as vehicle detectors, there is a case where useful travel time information that is useful can not be obtained unless it is densely installed on the route.
[0006]
In such a case, a method for calculating travel time information based on information obtained from toll road toll roads provided at a plurality of locations on the road has been studied. However, regarding this method, at the present time, the actual value of “how long it took for the vehicle that has arrived at the current destination” was calculated, and the required travel time information was created based on this value. It is. For this reason, changes in road conditions in the future are not taken into account since the travel time information is provided, and it is essentially ideal as travel time information, “travel time required to travel to the destination” In some cases, the accuracy was reduced.
[0007]
In addition, if the toll road's passage time is known on a toll road where a toll collection system is installed, the actual travel time is calculated from this passage time, and the travel time is a time transition based on this. A method has been proposed in which a pattern is created and stored, a pattern similar to the predicted date is searched from the stored pattern, and a travel time is predicted using the similar pattern. By using this method, prediction can be performed in consideration of what pattern change will occur in the future from the retrieved similar patterns in the past.
[0008]
However, since the prediction method based on the similar pattern depends on the searched similar pattern, it is important how to search for a pattern similar to the pattern on the prediction day.
[0009]
On the other hand, as a prediction method reflecting the traffic situation of the day, a method using cumulative traffic data has been proposed (for example, see Non-Patent Document 1).
[0010]
However, in the prediction method using the accumulated traffic volume data, the accuracy of the accumulated traffic volume prediction at a certain time ahead tends to deteriorate. In other words, the vehicle traveling speed largely depends on the intention of the vehicle driver, and it is difficult to determine how the driver responds to an event that occurs during traveling at a certain time or more, and the cumulative traffic volume The prediction accuracy tends to deteriorate. For this reason, there are cases where sufficient prediction accuracy cannot be obtained when the required travel time for a medium to long period including the time when the prediction accuracy of the accumulated traffic volume becomes low.
[0011]
[Non-Patent Document 1]
The Institute of Electrical Engineers of Japan Road Traffic Research Group RTA-01-14 “Time Estimate Model Using Pseudo Cumulative Traffic Volume” by Ueno, Oba, Kuwahara
[0012]
[Problems to be solved by the invention]
As described above, the method that can be used only on the route where the sensor is installed and the travel time information using data obtained from the tollgate so far do not take into account changes in the future road conditions. In some cases, it could not be said to be a typical travel time information. In particular, when the road conditions before and after traffic congestion change extremely, accurate travel time information cannot be obtained.
[0013]
When using similar pattern search that improves these, the prediction accuracy will depend on past accumulated data, so there is a problem that the prediction accuracy decreases if a pattern similar to the pattern on the prediction day cannot be searched. there were. Moreover, regarding the prediction using the accumulated traffic volume, although the traffic situation of the day can be reflected, the accuracy of the predicted value at a certain time ahead tends to deteriorate.
[0014]
An object of the present invention is to predict a travel time with high accuracy by combining a short-term travel time prediction using traffic data on the prediction day and a medium-long-term travel time prediction based on a similar pattern search. An object of the present invention is to provide a travel time predicting device that can perform the above-mentioned.
[0015]
[Means for Solving the Problems]
The travel time prediction apparatus according to the present invention collects and accumulates traffic information such as the number of vehicles that have passed through a plurality of data collection points arranged along the vehicle travel direction on the road and the passage time, and the travel time results. Based on the traffic data processing means that accumulates the travel time transition obtained from the past travel time actual value and the information collected and accumulated in this traffic data processing means, a predetermined time has elapsed from the present time Short-term travel time predicting means for predicting a short-term travel required time transition to the time, the predicted short-term travel required time transition, and the travel-time actual value obtained up to the present time obtained by the traffic data processing means The required travel time transition for the day is obtained from the time transition of the current time, and the similar travel required time transition similar to the travel time transition for the day is extracted from the accumulated travel time transition. , Characterized in that a required time estimating means for estimating the subsequent drive elapsed time using the similar drive elapsed time transition.
[0016]
The traffic data processing means can create a travel time transition using a road traffic flow simulator.
[0017]
In the present invention, the short-term required time prediction means includes cumulative traffic calculation means for obtaining cumulative traffic at each data collection point, and cumulative traffic prediction for predicting temporal transition of cumulative traffic after each point at each data collection point. And a time at which the accumulated traffic volume at the upstream data collection point that is the departure point becomes equal to the estimated cumulative traffic volume at the downstream data collection point that is the arrival point at the time when a predetermined time has elapsed since the present time. In addition, it is preferable to use one having a short duration travel time calculation means that takes a time difference between this time and the time at the arrival point as a travel time.
[0018]
Further, in the present invention, the short-term required time prediction means includes cumulative traffic calculation means for obtaining cumulative traffic volume at each data collection point, and cumulative traffic for predicting time transition of cumulative traffic volume at each data collection point after the present time. If there are multiple data collection points between the volume forecasting means and the departure point to the arrival point, the cumulative traffic volume at the arrival point at the time when a predetermined time has elapsed from the present time is obtained, and among the multiple upstream data collection points The time at the final upstream point where the accumulated traffic volume is the same as the cumulative traffic volume at the arrival point is obtained up to the present time, and the time difference between this time and the arrival point time is calculated from the final upstream point. The actual travel time value is used for the section from the final upstream point to the departure point, and the total value of these points is the travel point from the departure point to the arrival point. Or it may have a short-term traveling required time calculating means to time.
[0019]
Moreover, in this invention, what is obtained by integrating | accumulating the number of vehicles which passed each data collection point should just be used as a cumulative traffic calculation means.
[0020]
Further, in the present invention, the accumulated traffic volume calculation means is based on the accumulated traffic volume at the current time at the data collection point on the most downstream side of the target road, and travels in the section from the most downstream point to the upstream data collection point. Using the required time actual value and the travel required time actual value of each upstream section, based on the current time, the time sequentially following the travel time actual value of the corresponding section, the time at the corresponding upstream point The accumulated traffic volume at each point in time is calculated in a pseudo manner as the time when the accumulated traffic volume is reached, and the accumulated traffic volume predicting means is the time after the time corresponding to the pseudo accumulated traffic volume at each data collection point. It may have a function of predicting the time transition of the accumulated traffic.
[0021]
In the present invention, the short-term required time predicting means predicts the time transition of the required travel time between each data collection point after the present time from the actual travel time value up to the present time between the respective data collection points. Using the short-term travel time prediction means and the time transition of travel time in the section from the upstream data collection point that is the departure point to the data collection point that is the arrival point, A short duration travel time calculation means for obtaining a predicted travel time value for arrival and obtaining a travel time period from the departure point to the arrival point using a time that is retroactive from the arrival time as a departure time. It is good to use what has these.
[0022]
In the present invention, the short-term required time predicting means predicts the time transition of the required travel time between each data collection point after the present time from the actual travel time value up to the present time between the respective data collection points. When there are multiple data collection points from the departure point to the arrival point, the time transition of the required travel time in the section from the data collection point that is the arrival point to the upstream data collection point is calculated. Using the estimated travel time to arrive at the arrival point at the time when a predetermined time has elapsed from the present time, and the departure time of the upstream data collection point that is traced back from the arrival time by the estimated travel time estimated value. The arrival time at the upstream data collection point is used to determine the departure time of each upstream section until the departure point, using the travel time transition of the upstream section. Or it may have a short-term traveling required time calculating means for calculating a traveling time required to al arrival point.
[0023]
Further, in the present invention, traffic information such as the number of vehicles that have passed the data collection point and the passage time may be collected from the toll collection system at the toll gate on the toll road.
[0024]
In these inventions, the travel required time actual value is obtained from the traffic information from the plurality of data collection points collected by the traffic data processing means, and the travel required time transition is accumulated. The short-term required time predicting means predicts a short-term required travel time transition from the current time to a time when a predetermined time has elapsed. Then, from this short-term required time transition and the time transition of the travel required time actual value up to the present time, the required time prediction means obtains the travel required time transition of the day, and from the accumulated travel required time transition, A similar travel time transition similar to the travel time transition on the same day is extracted, and a subsequent travel time is predicted using this similar travel time transition.
[0025]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of a travel time prediction apparatus according to the present invention will be described with reference to the drawings. FIG. 1 shows the overall configuration of a travel time prediction apparatus according to an embodiment of the present invention.
[0026]
First, on highways, etc., cumulative traffic volume data is available at multiple points on the target road, or at the toll gate ID, entry / exit time, toll gate traffic at a toll gate such as a toll collection system. A road traffic system that can obtain data such as volume is adopted. The present invention is applied to such a road traffic system.
[0027]
In FIG. 1, reference numeral 1 denotes a toll road where a toll collection system is installed. A plurality of toll gates (corresponding to data collection points) 2 are installed on the toll road 1 along the vehicle traveling direction. These toll gates 2 serve as entrances and exits to toll roads and are connected to the toll collection system 3, and various traffic information such as toll collection information determined according to the travel distance is output to the toll collection system 3. Is done.
[0028]
A traffic data processing unit 4 includes a toll collection system data collection unit 41, a travel time actual value acquisition unit 5, and a pattern creation unit 6.
[0029]
The toll collection system data collecting means 41 collects traffic information such as the number of vehicles that have passed through the plurality of toll gates 2, the passage time, and the vehicle type from the toll collection system 3. The required travel time actual value acquisition means 5 includes a toll collection system data storage means 51 for storing the collected traffic information, and a required travel time for determining the actual required travel time value between the toll stations based on the collected traffic information. And actual value calculation means 52.
[0030]
The pattern creating unit 6 includes a travel required time pattern creating unit 61 and a travel required time pattern accumulating unit 62. The required travel time pattern creating means 61 obtains a required travel time pattern (corresponding to the required travel time transition) at a constant cycle (for example, a daily cycle) based on the past actual travel time values, and the required travel time. It accumulates in the pattern accumulation means 62.
[0031]
Reference numeral 7 denotes short-term required time predicting means for predicting a short-term required travel time transition from a current time to a time when a predetermined time has elapsed based on information collected and accumulated in the traffic data processing means 4. This short period required time predicting means 7 includes cumulative traffic amount calculating means 71, cumulative traffic amount predicting means 72, and short period required travel time calculating means 73.
[0032]
The accumulated traffic calculation means 71 obtains the accumulated traffic at each toll gate (data collection point) 2. The cumulative traffic volume is a cumulative value of the number of passing vehicles at each tollgate 2 and increases with time.
[0033]
The cumulative traffic volume predicting means 72 predicts the temporal transition of the cumulative traffic volume after the present time at each data collection point. That is, if the accumulated traffic volume is calculated using the actual value of the number of passing vehicles at each tollgate 2, the actual accumulated traffic volume value up to the present time is obtained as shown by the black circles of curves a and b in FIG. . Cumulative traffic volume (white circles) after the current time is calculated using methods such as the rate of change in cumulative traffic volume up to the current time, using autoregressive models, using polynomial approximation models, and using neural networks. Can be sought.
[0034]
The short period travel time calculation means 73 obtains the travel time in a relatively short period from the present time to a predetermined time ahead from the time transition of the accumulated traffic volume. That is, as shown in FIG. 3, using the accumulated traffic curve a at the upstream toll booth A and the accumulated traffic curve b at the downstream toll booth B, the travel time between the toll booths A and B is calculated. Ask. In the example of FIG. 3, first, the accumulated traffic b1 at the downstream toll booth B (arrival point) at the time t + 2Δt when a predetermined time has elapsed from the current time t is obtained from the curve b. Next, using curve a, the time when the accumulated traffic volume at the upstream toll booth A (departure point) becomes equal to the estimated accumulated traffic volume b1 at the downstream toll booth B (t-4Δt and t in the example in the figure). Between -5Δt). Then, the time difference between the time of the toll booth A and the time of the toll booth B that becomes the cumulative traffic volume predicted value b1 is set as the required travel time.
[0035]
Reference numeral 8 denotes required time prediction means, which includes a required travel time pattern creation means 81 on the day, a similar required travel time pattern extraction means 82, and a required travel time prediction calculation means 83.
[0036]
The required travel time pattern creating means 81 on the day is a time transition of the short travel required time obtained by the short time required time predicting means 7 and the actual travel required time value obtained up to the present time obtained by the traffic data processing means 4. The time required travel pattern on the day (corresponding to the time required travel on the current day) is determined from the time transition of the current time.
[0037]
The similar travel required time pattern extracting means 82 is similar to the required travel time pattern of the day determined from the past required travel time pattern stored in the travel required time pattern storage means 62 as described above. Extract the pattern.
[0038]
The required travel time calculation means 83 predicts the required travel time thereafter using the extracted similar required travel time pattern.
[0039]
Next, the operation will be described with reference to the flowchart of FIG.
[0040]
First, fee collection system data obtained by the fee collection system 3, for example, entrance passage time, exit passage time, vehicle type, and the like are acquired by the fee collection system data acquisition means 41 (S101). As a toll collection system, if it is a current toll collection system using a magnetic toll ticket, data obtained from a magnetic toll ticket may be used, and if it is an ETC (non-stop automatic toll collection system), it can be obtained from an ETC. Can be used.
[0041]
Next, the fee collection system data acquired by the fee collection system data acquisition unit 41 is stored in the fee collection system data storage unit 51 (S102). The travel required time actual value calculation means 52 uses the charge collection system data stored in the charge collection system data storage means 51 to calculate the required travel time actual value between the toll gates (S103).
[0042]
The actual travel time actual value is calculated based on the entrance toll gate passage time and the exit toll gate passage time acquired by the toll collection system data acquisition means 41. The following formula is used as the simplest method.
[0043]
Travel time actual value = Exit toll gate passage time-Entrance toll gate passage time
The travel required time actual value obtained by the calculation means 52 is stored in the data storage means 51 together with traffic information.
[0044]
Note that the actual travel time value between the toll booths obtained by the above calculation is affected by the acceleration / deceleration near the entrance / exit. For this reason, since there may be a delay from the required travel time when traveling only on the main line from upstream to downstream, correction such as subtraction of the delay time calculated based on statistical values or the like may be performed.
[0045]
Next, the required travel time pattern creating means 61 uses the actual travel required time value stored in the toll collection system data storage means 51 for every predetermined period (for example, one day) (S104: YES). The pattern is calculated (S105). The travel required time pattern is obtained by calculating a temporal transition of travel required time data over a predetermined cycle (for example, 1 day) at certain time intervals (for example, every 5 minutes). The simplest method for creating a pattern is to use an average of the obtained data.
[0046]
However, toll roads often have a service area and a parking area, and there are unique data such as a car that has stopped for a long time for a break. As a method for removing these unique data and creating a pattern, the methods disclosed in Japanese Patent Application Laid-Open Nos. 2000-259983 and 2000-285372 proposed by the applicant may be used. Furthermore, when creating the travel time pattern, as shown in Japanese Patent Application No. 2002-74784 proposed by the applicant, filtering may be performed to remove small changes in the travel time pattern or to set a maximum value. In addition, the travel time pattern may be normalized.
[0047]
It should be noted that the travel time pattern may not be (sufficiently) available if the actual system is introduced before the road is in operation or shortly after the start of operation. In this case, a travel time pattern may be created virtually using a virtual traffic flow using a road traffic flow simulator. In this case, if the actual travel time value can be obtained as much as possible, the characteristics of the road traffic flow are analyzed based on this and reflected in the parameters of the road traffic flow simulator. Can be created.
[0048]
Thus, the required travel time pattern created by the required travel time pattern creating means 61 is stored in the required travel time pattern accumulating means 62 (S106). As a storage method, the data may be stored in time series in a computer memory or a data storage medium such as a hard disk.
[0049]
Further, the accumulated traffic calculation means 71 uses the toll collection system data (traffic volume, travel time, etc.) accumulated in the toll collection system data accumulation means 51 to create the accumulated traffic (S107). The accumulated traffic volume can be created by integrating the traffic volume passing through the entrance / exit toll gate included in the toll collection system data. That is, here, the actual value of the accumulated traffic volume at each tollgate up to the present time is calculated.
[0050]
Next, the cumulative traffic volume predicting means 72 uses the cumulative traffic volume actual value obtained up to the current time obtained by the cumulative traffic volume calculating means 71 to predict the cumulative traffic volume after the current time (S108).
[0051]
Next, by using the cumulative traffic volume predicted value calculated by the cumulative traffic volume predicting means 72 by the short period travel time calculating means 73, the required travel time when arriving at the destination after a certain time from the current time is calculated. The calculation is performed using the relationship between the accumulated traffic volume and the travel time described in FIG. 3 (S109). That is, the accumulated traffic volume of the destination B at a certain time after the current time is examined, the time when the accumulated traffic volume of the departure place A is the same value is extracted, and the difference between these times is obtained, The estimated travel time from the departure point A to the downstream destination B is calculated.
[0052]
Next, the short-term required travel time prediction calculated by the short-term required travel time calculation means 71 is calculated by the short-term required travel time calculation means 71 to the current travel required time actual value read from the toll collection system data storage means 81 by the required travel time pattern creating means 81. The value is combined and corrected to create a travel time pattern (corresponding to the travel time transition on that day) up to a certain time ahead on the predicted day (S110).
[0053]
Next, the similar travel required time pattern extracting unit 82 calculates the travel required time pattern most similar to the travel required time pattern up to a time ahead of the predicted day calculated by the current day travel required time pattern creating unit 81. Extracted from the past required travel time pattern stored in the time pattern storage means 62 (S111). For extracting the most similar travel time pattern, for example, a method based on the calculation of the square error as described in JP-A-2001-126180 may be used.
[0054]
When using the travel time data for the current day, filtering the available data (actual values and predicted values) for the current day will reduce the adverse effects of the predicted data even if there is a slight vibration. (See Japanese Patent Application No. 2002-74784). Further, if the required travel time pattern created by the required travel time pattern creating means 61 and the pattern of the day are normalized based on the maximum value, even if a relatively small required travel time pattern is accumulated, it is accurate. The travel time can be predicted (see Japanese Patent Application No. 2002-74784). By such a process, the required travel time pattern most similar to the travel required time data on the predicted day can be extracted.
[0055]
Finally, the required travel time calculation means 83 predicts the required travel time when the departure point is departed at the current time using a similar required travel time pattern that is most similar to the travel required time data on the predicted day (S113). . In the extracted similar travel time pattern, since the travel time data is arranged corresponding to the time passing through the exit toll gate, this is rearranged to correspond to the time passing through the entrance toll gate, This is used to calculate a predicted travel time value when traveling in the prediction target section. The simplest way to calculate the estimated travel time is to use the travel time pattern rearranged in the order of the entrance toll gate passage time, and use the travel time at the time corresponding to the predicted time as the predicted value. There is a method to use.
[0056]
In addition, when the similarity of the extracted similar travel required time pattern is low, a predicted travel required time that may not be actually calculated may be calculated, but if you look at the time transition of the actual travel required actual value, Since the limit of the change is known, by correcting the travel time using this limit, it is possible to correct an impossible prediction result from the temporal transition of the actual travel time (Japanese Patent Application No. 2002-74784) reference).
[0057]
With the above operation, the estimated travel time estimated value is calculated.
[0058]
Thus, the predicted value of the required travel time can be calculated on the toll road where the toll collection system is installed. In addition, the transition of short-term travel time up to the near future is predicted with high accuracy using the accumulated traffic volume based on the traffic data on the predicted day, and this predicted value is added to the travel time pattern of the current day. By performing a similar pattern search, the travel time can be predicted with high accuracy while reflecting the traffic situation of the day.
[0059]
In the above embodiment, the actual accumulated traffic volume value up to the present time is obtained at each toll gate. However, depending on the structure of the toll gate, the actual accumulated traffic volume value may not be obtained. Use cumulative traffic volume. For example, for entrance toll gates, there are cases where the entrance toll gates for up and down are the same, and it is not always possible to obtain the correct accumulated traffic volume by simply adding the number of passing cars. In this case, the cumulative traffic volume near each toll gate is calculated in a pseudo manner from the outflow traffic at the exit toll gate at the downstream end and the travel time between the toll gates (RTA-01- 14 (See Ueno, Oba, and Kuwabara) “Time required model using simulated cumulative traffic”.
[0060]
In other words, the outflow traffic volume at the tollgate at the most downstream end of the toll road is a correct value without error, and this outflow traffic volume is taken as the accumulated traffic volume at the most downstream end, and the accumulated traffic volume at each upstream tollgate is simulated. Ask for.
[0061]
In this case, the accumulated traffic volume calculation means 71 obtains the accumulated traffic volume actual value at the current time at the tollgate at the most downstream of the target road and the actual travel time actual value between the respective toll stations upstream from this. . The accumulated traffic volume value at the most downstream is considered to have passed through the toll booth immediately upstream at the time required for traveling in the section with the upstream toll booth. Estimated as the cumulative traffic volume at the time required to go back. Hereinafter, similarly, for each toll station on the upstream side, the actual travel time value of each section on the upstream side is used, and as a cumulative traffic volume at the time traced back from the time actual value for the corresponding station from the downstream time, Determine.
[0062]
For upstream toll stations other than the most downstream toll gate, the accumulated traffic volume from the time when the accumulated traffic volume at the most downstream toll gate at the current time is the same as the accumulated traffic volume (different for each toll gate) to the current time May be estimated together with the prediction of the previous accumulated traffic volume from the current time by the accumulated traffic volume predicting means 72.
[0063]
In other words, the cumulative traffic volume predicting means 72 has a function of estimating the time transition of the cumulative traffic volume after the time corresponding to the pseudo cumulative traffic volume at each data collection point. Note that the obtained accumulated traffic volume may be filtered by the same method as that for creating the required travel time pattern to suppress the influence of small variations.
[0064]
When such pseudo cumulative traffic volume is used, for upstream toll stations other than the most downstream toll gate, the current time from the time when the accumulated traffic volume is the same as the accumulated traffic volume at the most downstream toll gate at the current time. The accumulated traffic volume up to is an estimated value, and may be calculated without using the travel required time actual value data, such as when the accumulated traffic volume predicting means 72 predicts it together with the portion ahead of the current time.
[0065]
Therefore, for the portion past the current time, the travel required time may be predicted using the travel required time actual value data of each section without using the accumulated traffic volume.
[0066]
For example, in FIG. 4, curve a is the cumulative traffic volume at the upstream toll booth A at the departure point, curve c is the cumulative traffic volume at the toll booth C at the intermediate point, and curve b is the cumulative traffic volume at the downstream toll booth B as the arrival point. It is a curve. The other curves d and e are also cumulative traffic volume curves of the toll gates present in the middle part.
[0067]
When using the accumulated traffic volume at each toll booth to determine the required travel time for a short period of time ahead of the current time, first, using the accumulated traffic curve b, a certain time after the current time. The cumulative traffic volume predicted value b1 of the destination (arrival point) B at the time is examined. Next, the upstream toll gate where the estimated cumulative traffic volume reaches the latest value b1 until the current time is examined using each cumulative traffic curve. In the example shown in the figure, the toll gate C reaches the accumulated traffic volume b1 the latest. Therefore, using the cumulative traffic curve c, the time to reach the cumulative traffic b1 at the toll gate C is obtained, and this is used as the departure time, and the toll gate when arriving at the toll gate B at a certain time after the current time. The required travel time X between C and B is obtained.
[0068]
Next, the required travel time Y from the toll booth D that arrives at the toll booth C at the departure time at the toll booth C is obtained. Since the vehicle travels before the current time during this period D to C, a travel time actual value is obtained, and is obtained by linear interpolation or the like using the actual value. Similarly, the required travel time Z in the sections A to C is obtained from the actual value by going back to the departure place A. As a result, the travel time using the actual value is obtained between the toll stations A to C that traveled before the current time, and the cumulative traffic curve b between the toll stations C and B including the travel after the current time. , C is used to obtain the required travel time, and the total of these is the predicted travel time between the toll stations A and B.
[0069]
In summary, when there are a plurality of tollgates (data collection points) from the departure point A to the arrival point B, the accumulated traffic volume b1 at the arrival point B at the time when a predetermined time has elapsed from the present time is obtained, and a plurality of upstream sides Among the data collection points, the time at the last upstream point C that has reached the same cumulative traffic volume as the cumulative traffic amount b1 of the arrival point B until the present time is obtained, and the time difference between this time and the time of the arrival point Is the required travel time from the final upstream point C to the arrival point B, the section from the final upstream point C to the departure point A uses the actual travel time value, and these total values are calculated from the departure point A to the arrival point B. Estimated as the required travel time.
[0070]
In this way, when predicting the short duration travel time after the lapse of a predetermined time from the current time, when the section traveled before the present time is included, the actual travel time is used to calculate the final travel time. Prediction accuracy can be increased.
[0071]
FIG. 5 shows a configuration of a required travel time prediction apparatus according to another embodiment of the present invention.
[0072]
This travel time prediction device is different from the one shown in FIG. 1 in the short time required time prediction means 7, and the other parts are the same as those in FIG. This short period required time predicting means 7 includes a short period required travel time predicting means 74 and a short period required travel time calculating means 75.
[0073]
The short-term travel required time predicting means 74 predicts the time transition of the travel required time between each tollgate after the current time from the actual travel required time value between the toll gates at the current time.
[0074]
The short-term travel time calculation means 75 uses the time transition of travel time in the section from the upstream toll booth as the departure point to the toll gate as the arrival point, and arrives at the arrival point at the time when a predetermined time has elapsed from the present time. The estimated travel time is calculated. Then, the time required for traveling from the departure point to the arrival point is obtained with the departure time as the time that is obtained from the estimated time required for traveling.
[0075]
Next, the operation will be described with reference to the flowchart of FIG.
[0076]
The operations from Steps S201 to S206 in FIG. 6 correspond to Steps S101 to S106 in FIG. 2 and are the same operations.
[0077]
That is, information on each vehicle that has passed through the toll booth 2 of the target toll road 1 is available from the toll collection system 3, the toll collection system data is acquired by the toll collection system data acquisition means 4, and the toll collection system The data is accumulated in the data accumulating means 51. Based on this data, the required traveling time actual value calculating means 52 calculates the actual traveling required time value between the toll stations, and stores it in the toll collection system data accumulating means 51. The required travel time pattern creating means 61 calculates the required travel time pattern from the actual travel required time value for each cycle and stores it in the required travel time pattern storage means 61 in the embodiment described with reference to FIGS. Is the same.
[0078]
In this embodiment, the toll collection system data (traffic volume, travel time, etc.) stored in the toll collection system data storage means 51 is used by the short period travel time prediction means 74 of the short period required time prediction means 7. Then, the transition of the required travel time for a short period between the toll stations until the time when the predetermined time has elapsed is predicted (S208). When the travel required time actual value of the entire process of the target section is obtained, the travel required time of the target section may be directly predicted.
[0079]
To estimate the required travel time, use the actual travel time value up to the present time, use the rate of change of the required travel time, use an autoregressive model, use a polynomial approximation model, or use a neural network , Etc. are used.
[0080]
Next, the short-term travel required time calculation means 75 uses the predicted travel-time required value calculated by the short-term travel required time prediction means 74, and the short-term when the vehicle arrives at the destination after a certain time from the current time. The required traveling time is calculated (S209).
[0081]
Here, when the travel required time between each toll booth is predicted by the short period travel required time predicting means 74, the travel time required for each section may be traced back from the destination side to the departure point.
[0082]
For example, assuming that the toll gates are arranged in the order of A, D, C, B from the upstream side of the target road, FIG. 7 shows the transition of the travel time of each toll gate section, and the curve ad Represents the transition of the required travel time in the upstream sections A to D, the curve cb represents the transition of the required travel time in the downstream sections C to B, and the curve dc represents the transition of the required travel time in the midstream sections D to C.
[0083]
In such a state, when the required travel time between the toll booths A and B after the current time is calculated, first, the curve cb is used when arriving at the toll booth B at a time (t + 2Δt) after a predetermined time from the current time. The required travel time X (predicted value) between the sections C to B is obtained. Next, the departure time from the toll gate C, which is the required travel time X from the arrival time, is the arrival time at the toll gate C in the upstream section, and the required travel time between the sections D to C by the curve dc. Y (predicted value) is obtained. In the same manner, the departure time from the toll booth D, which is the required travel time Y from the arrival time at the toll booth C, is set as the arrival time at the toll booth D in the further upstream section by the curve ad. A required travel time Z between D is obtained (actual value in the example in the figure). The time required for traveling Z from the arrival time at the toll booth D is taken as the departure time for the toll booth A as the departure point, and the time difference from the departure time at the toll booth A to the arrival time at the toll booth B (X + Y + Z) is predicted as the required travel time between the toll gates A to B after a predetermined time.
[0084]
That is, the required travel time is calculated by linear interpolation or the like based on the predicted travel time from the toll stations C to B immediately upstream of the toll booth B. Next, the required travel time from the toll booth D that arrives at C at the departure time of C is calculated by linear interpolation or the like based on the predicted travel time between D and C. Further, the required travel time from the toll booth A (departure point) on the upstream side, which arrives at D at the departure time of D, is the required travel time between A and D (the arrival time at D in the example in the figure). Is a past time from the current time, and is calculated by linear interpolation or the like based on the measured value). As a result, the required travel time for the target sections A to B is obtained.
[0085]
The above explanation is a case where there are a plurality of tollgates between the target sections A to B, but when the required travel time of the target sections A to B can be directly predicted by the short distance travel required time prediction means 74, Based on the estimated travel time estimate value, the travel time required to arrive at the destination after a certain time from the current time is calculated by linear interpolation or the like.
[0086]
That is, using the time transition of the travel required time directly obtained for the section from the departure point A to the arrival point B, the travel required time predicted value arriving at the arrival point B at the time when a predetermined time has elapsed from the present time is obtained, The required travel time from the departure point A to the arrival point B is obtained by using the time that has been obtained from the arrival time and the predicted travel time estimated value as the departure time.
[0087]
Thereafter, as in the embodiment described with reference to FIGS. 1 and 2, the required travel time actual value up to the present time read from the toll collection system data storage means 51 and the short-term travel required time pattern creating means 81. The required travel time pattern up to a certain time ahead of the predicted day is created using the short-term travel required time prediction value calculated by the required travel time calculation means 75 (S210), and similar travel required time pattern extraction means A similar pattern search is performed at 82 (S211), and the travel time calculation means 83 predicts the travel time required when departing from the departure place at the current time (S210).
[0088]
In this way, even if the accumulated traffic volume is not calculated for each toll booth, if the actual travel time between toll gates is calculated, the transition of the travel time to the near future It can be used and predicted with high accuracy. For this reason, the predicted value is added to the travel time pattern of the current day, and the similar pattern search is performed, thereby reflecting the traffic situation of the current day and preventing the deterioration of the prediction accuracy.
[0089]
Each of the above embodiments relates to a toll road to which the toll collection system is applied, but can be applied to any system as long as data on the time of passage at the entrance / exit toll gate and the vehicle type are available. Therefore, by replacing the toll collection system with ETC (non-stop toll collection system), it can be easily applied to ETC. Furthermore, the present invention can be applied to an AVI system (automatic vehicle recognition system) as long as the passage time at any two points is available. Or, in the future, infrastructure equipment that can obtain the actual travel time value, such as a road-to-vehicle communication system, has been developed and can be applied if the travel time between any two points is available.
[0090]
FIG. 8 is a diagram for explaining the effect of the present invention, and in each case, the transition of the required travel time is represented by a curve. A curve α in FIG. 6A shows the actual value up to the predicted time of the day. A curve β in FIG. 5B shows a similar pattern extracted by performing a similar pattern search using the actual value α up to the predicted time. As is clear from the figure, there is a considerable difference with the passage of time when compared with the actual value α1 after the predicted time.
[0091]
A curve γ in FIG. 5C shows a case where a transition after the prediction time is predicted by an autoregressive model or the like using the actual value α up to the prediction time. As is clear from the figure, the prediction accuracy is almost the same as the actual value α1 until a time after a predetermined time has elapsed from the predicted time, but after the predetermined time has elapsed, the actual value α1 deviates. Therefore, when prediction is performed using the prediction curve γ, a prediction error ahead for a long time after a predetermined time has elapsed.
[0092]
FIG. 4D shows the prediction characteristics obtained by the present invention, and the curve δ is a similar pattern based on the current day pattern that combines the actual value up to the prediction time and the short-term prediction value up to the time after a certain time has elapsed. It is a similar pattern obtained by searching. As is apparent from the figure, it has been in a state close to the actual value α1 over a medium to long period after a predetermined time has elapsed. Therefore, if this prediction curve (similar pattern) δ is used, the prediction accuracy is improved over a medium to long period.
[0093]
【The invention's effect】
According to the present invention, since the required travel time for a short time using the traffic data on the prediction day is combined with the predicted required travel time for the medium to long term by the similar pattern search, the required travel time with high accuracy is predicted. This can reduce the driver's stress and ensure the reliability of the travel time information.
[Brief description of the drawings]
FIG. 1 is a system configuration diagram showing an embodiment of a travel time estimation device according to the present invention.
FIG. 2 is a flowchart for explaining the operation of the embodiment.
FIG. 3 is a characteristic diagram for explaining a travel time prediction method according to the embodiment.
FIG. 4 is a characteristic diagram for explaining another traveling time prediction method according to the embodiment.
FIG. 5 is a system configuration diagram showing another embodiment of a travel time estimation device according to the present invention.
FIG. 6 is a flowchart for explaining the operation of another embodiment;
FIG. 7 is a characteristic diagram for explaining a travel time prediction method according to another embodiment.
FIG. 8 is a characteristic diagram for explaining the effect of the present invention.
[Explanation of symbols]
1 road
2 Data collection points
4 Traffic data processing means
5 Traveling time actual value acquisition means
51 Data storage means
52 Traveling time actual value calculation means
6 Pattern creation means
61 Traveling time pattern creation means
62 Traveling time pattern storage means
7 Short-term required time prediction means
71 Cumulative traffic calculation means
72 Cumulative traffic prediction means
73 Short-term travel time calculation means
74 Short-term travel time prediction means
75 Short term travel time calculation means
8 Time required calculation means
81 Traveling time pattern creation means on the day
82 Similar travel time pattern extraction means
83 Traveling time calculation means

Claims (9)

Collects and accumulates traffic information such as the number of vehicles that have passed through multiple data collection points arranged along the road's vehicle travel direction and the time of passage to obtain the actual travel time value and past travel time Traffic data processing means for accumulating the travel time transition obtained from the actual value of
Based on the information collected and accumulated in the traffic data processing means, the short-term required time predicting means for predicting the short-term travel required time transition from the current time to the time when a predetermined time has elapsed,
From the predicted short-term travel time transition and the time transition of the travel time actual value obtained up to the present time obtained by the traffic data processing means, the travel time transition on the day is obtained, and the accumulated travel time transition From the above, a required travel time prediction means for extracting a similar travel time transition similar to the travel time transition on the same day and predicting a travel time thereafter using this similar travel time transition,
A travel time prediction apparatus comprising: The travel time prediction device according to claim 1, wherein the traffic data processing means creates travel time transition using a road traffic flow simulator. The short-term required time prediction means is
Cumulative traffic calculation means for determining cumulative traffic at each data collection point;
Cumulative traffic prediction means for predicting the time transition of the cumulative traffic after each point at each data collection point;
Find the time when the accumulated traffic volume at the upstream data collection point that is the departure point is equal to the estimated cumulative traffic volume at the downstream data collection point that is the arrival point at the time when a predetermined time has elapsed from the current time. And a short-term required travel time calculation means that sets the time difference between the arrival point and the time as travel required time,
The travel time prediction apparatus according to claim 1, wherein The short-term required time prediction means is
Cumulative traffic calculation means for determining cumulative traffic at each data collection point;
Cumulative traffic prediction means for predicting the time transition of the cumulative traffic after each point at each data collection point;
If there are multiple data collection points between the departure point and the arrival point, calculate the cumulative traffic volume at the arrival point at the time when a predetermined time has elapsed from the current point. Find the time at the last upstream point that has the same cumulative traffic volume as the cumulative traffic volume at the arrival point, and calculate the time difference between this time and the arrival point time from the final upstream point to the arrival point. And, the section from the final upstream point to the departure point uses the travel required time actual value, and the short period travel required time calculation means that sets the total required value from the departure point to the arrival point,
The travel time prediction apparatus according to claim 1, wherein The travel time prediction apparatus according to claim 3 or 4, wherein the accumulated traffic volume calculation means is obtained by integrating the number of vehicles that have passed through each data collection point. The accumulated traffic volume calculation means is based on the accumulated traffic volume at the current time at the data collection point on the most downstream side of the target road, and the actual travel time required value for the section from the most downstream point to the upstream data collection point, And, using the actual travel time value of each upstream section, based on the current time, the time that was sequentially traced to the actual travel time value of the corresponding section reached the cumulative traffic volume at the corresponding upstream point. Calculate the cumulative traffic volume at each point in time as a time,
The accumulated traffic volume predicting means has a function of predicting a temporal transition of accumulated traffic volume after each time corresponding to the pseudo accumulated traffic volume at each data collection point. The travel time prediction device according to claim 1. The short-term required time prediction means is
A short-term required travel time prediction means for predicting a time transition of travel required time between each data collection point from the actual travel time actual value between each data collection point and the present time;
Using the time transition of the travel time of the section from the upstream data collection point that is the departure point to the data collection point that is the arrival point, obtain the estimated travel time that will arrive at the arrival point at the time when a predetermined time has elapsed from the current time A short-term required travel time calculation means for determining a required travel time from the departure point to the arrival point, with the departure time as a departure time estimated from the estimated required travel time value from the arrival time;
The travel time prediction apparatus according to claim 1, wherein The short-term required time prediction means is
A short-term required travel time prediction means for predicting a time transition of travel required time between each data collection point from the actual travel time actual value between each data collection point and the present time;
When there are multiple data collection points from the departure point to the arrival point, the time when a predetermined time has elapsed from the present time using the time transition of the travel time of the section from the data collection point that is the arrival point to the upstream data collection point The estimated travel time required to arrive at the arrival point is obtained, and the departure time of the upstream data collection point retroactive from the arrival time by the calculated travel time prediction value is determined as the arrival time at the upstream data collection point. Using the travel time transition of the upstream section, sequentially determining the departure time of each upstream section until the departure point, and a short period travel time calculation means for obtaining the travel time from the departure point to the arrival point,
The travel time prediction apparatus according to claim 1, wherein The travel time prediction apparatus according to claim 1, wherein traffic information such as the number of vehicles that have passed the data collection point and the passage time is collected from a toll collection system at a toll gate on a toll road.

Images