JP2002260142A - Predicting method for traveling time - Google Patents

Abstract

PROBLEM TO BE SOLVED: To exactly predict traveling time for later that day, with consideration given to the previous traveling time data of the same section of a road and to the transition of traveling time up until the day immediately before leaving. SOLUTION: The traveling time data up until the current time on one or more section(s) of the road including the section X which is the subject of traveling time measurement is actually measured (Figure 7 (b)), and the previous traveling time memory table (Figure 7 (a)) is specified which materializes the traveling time pattern of the same time of the day as the actually measured traveling time data pattern, or a similar traveling time pattern of the similar time of the day. The memorized traveling time for similar traveling time pattern of N hours ahead at the section X of the road is determined as predicted traveling time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for estimating a travel time of a road section.

[0002]

2. Description of the Related Art In the field of road traffic information processing, it is important to predict the time required for a vehicle to travel on a road ("running time"). Using this travel time, an optimal route to be taken from one point (departure point) to another point (destination point) can be accurately calculated and provided to the driver. Further, the driver can know the travel time and the arrival time while arriving at the destination.

[0003]

By the way, in order to predict the travel time, the past travel time of the road is actually measured,
A technique of estimating a future traveling time using the statistical data is used. For example, there is a method in which travel time data of the same month, the same day, the same day of the week, the same time zone, or the like in the past is used as the travel time of the previous time as it is. However, the travel time of the road on that day does not always reproduce a value according to past data.

[0004] Therefore, the inventor has noted that it is important to refer to the past data as well as the transition of the travel time immediately before the day in order to accurately predict the travel time of the day. Further, even if the predicted value of the traveling time can be obtained, if the reliability of the prediction is known, for example, when the optimal route is calculated, it is possible to know the degree of the optimality of the route. Further, it is possible to evaluate an error included in the travel time required to reach the destination.

An object of the present invention is to realize a traveling time prediction method capable of accurately predicting the traveling time of the day by taking into account the transition of the traveling time until immediately before the day together with the past data. . Another object of the present invention is to realize a traveling time prediction method capable of evaluating the reliability of the predicted traveling time.

[0006]

Means for Solving the Problems and Effects of the Invention (1)
The travel time prediction method according to the present invention includes a road section X whose travel time is to be predicted, and a time N from the current time to the prediction time.
Is determined, and the driving time data up to the current time of one or more road sections including the road section X are collected, and from the driving time storage tables collected in the past in the road section,
A pattern similar to the collected travel time data and the travel time is searched for, and the travel time storage value of the road section X N hours ahead of the pattern similar to the travel time in the travel time storage table is calculated as a travel time prediction value. (Claim 1).

[0007] According to the above method, the travel time data of one or more road sections including the road section X whose travel time is to be predicted is measured up to the current time, and the pattern of the measured travel time data is measured. For example, a past travel time storage table that realizes a pattern similar to the time distribution pattern of the travel time and the increase / decrease time distribution pattern of the travel time is specified. Then, the travel time storage value of the road section X N hours ahead of the pattern having a similar travel time in the travel time storage table is set as the travel time prediction value.

According to this method, the running time of the current day can be accurately predicted in consideration of the transition of the running time immediately before the current day to the past statistical data. It is preferable that the range of time for collecting the travel time data is from a time before a predetermined time in the past to the current time (claim 2). Even if the range of time for collecting travel time data is taken from a long time ago,
It cannot be said that accurate predictions can be made. In reality, it is sufficient to collect data of the day (for example, data from the morning of the day).

[0009] A plurality of similar pattern candidates in the travel time storage table are searched, and the mode value of the travel time predicted value obtained for each of them can be determined as the travel time predicted value. . According to this method,
The travel time prediction value can be determined more accurately. (2) In addition, the travel time prediction method of the present invention obtains a travel time prediction value of the road section X whose travel time is to be predicted based on past statistical travel time data, and calculates the travel time prediction value. This is a method for obtaining reliability (claim 4).

Although various methods are known for obtaining the predicted travel time of the road section X, it is considered that there was no idea to obtain the reliability of the predicted travel time. According to the present invention, by determining the reliability of the travel time prediction value,
The utility value of the travel time can be increased. The travel time prediction method according to the present invention determines a road section X for which a travel time is to be predicted and a time N from the current time to the prediction time, and determines the current time of one or more road sections including the road section X. The travel time data up to and including the travel time data collected in the past in the road section are searched for a plurality of candidates of patterns similar to the collected travel time data and the travel time. The travel time storage value of the road section X N hours ahead of the pattern having a similar travel time is used as a candidate value of the travel time prediction value, and the mode value of the candidate value of the travel time prediction value is calculated as the travel time prediction value. And determining the reliability of the travel time prediction value based on the spread of the distribution of the candidate values of the travel time prediction value (claim 5).

This method specifically defines the method according to claim 4. A plurality of similar pattern candidates in the travel time storage table are searched, and the mode value of the travel time predicted value obtained for each is determined as the travel time predicted value, and the distribution of the candidate values of the travel time predicted value is determined. The reliability of the predicted travel time is determined based on the spread of the travel time.
If the distribution of candidate values is wide, the reliability of the predicted travel time is determined to be low, and if the distribution of candidate values is concentrated, the reliability of the predicted travel time is determined to be high.

[0012]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings. FIG. 1 is a perspective view showing a road network including an intersection, a beacon communication device 23, an ultrasonic vehicle sensor 24, and a camera-type license plate detection device 25 installed on the road network. The vehicle is traveling on the left side, the beacon communication device 23 is mainly installed at the intersection exit, the vehicle detector 24 is mainly installed at the intersection entrance, and the license plate detection device 25 is installed at a specific intersection.

FIG. 2 is a detailed diagram of the beacon communication device 23. The beacon communication device 23 has a control device 32 attached to a pole 31 on the side of the road, and an infrared light emitter / receiver 33 fixed to a beam above the road. Emitter / receiver 33
Is in the immediate lane. The beacon communication device 23 transmits data of travel time, traffic congestion information, and regulation information when a vehicle equipped with the on-vehicle communication device 34 passes right below. Receives data such as a point at which transmission / reception was previously performed with the beacon communication device and a travel time from that point to a point at which transmission / reception with this time beacon is performed. This in-vehicle communication device 34
The data transmitted from the to the beacon communication device 23 is referred to as “uplink data”.

Here, a method of obtaining road travel time data is known. For example, a vehicle can be specified based on the uplink data and the license plate detecting device 25.
3, the time required to pass another beacon communication device 23 may be obtained. FIG. 3 is a configuration diagram of an entire system including the traveling time prediction device 13. The travel time prediction device 13 is connected to the data tally unit 12. The data tallying unit 12 is connected to the beacon communication device 23, the ultrasonic vehicle sensor 24, and the license plate detection device 25 through a line. The traveling time prediction device 13 includes a computer, a memory, an input / output device, and the like, and performs the following functions.

FIG. 4 is a functional block diagram of the traveling time prediction device 13. The latest travel time data of each road section is input from the data aggregation unit 12 to the travel time prediction device 13. The input travel time data is temporarily stored in the storage unit 14 and supplied to the fluctuation pattern storage unit 17. The variation pattern storage unit 17 accumulates the traveling time data for a plurality of past days and outputs it as a traveling time storage table on a daily basis. The travel time data stored in the storage unit 14 is sent to the evaluation unit 15, where the evaluation unit 15
Comparison and evaluation with the running time storage table stored in the variation pattern storage unit 17 are performed.

The evaluation result of the evaluation section 15 is sent to a traveling time determination section 16. The traveling time determination unit 16 determines and outputs a predicted traveling time based on the evaluation result. Further, the evaluation result by the evaluation unit 15 and the travel time prediction value are sent to the error calculation unit 18, and the error calculation unit 18 calculates the calculation error of the travel time prediction value. FIG. 5 is a diagram showing a specific example of the travel time storage table. The travel time storage table is
Road section 1 from midnight to 12:00 pm every day
The travel time data measured every 5 minutes of 2, 3,... Is stored. The road sections 1, 2, 3,... Are road sections in the target area of the travel time prediction service. The pattern number P is a number representing a day.

FIG. 6 is a flowchart for explaining a travel time prediction method performed by the evaluation unit 15 and the travel time determination unit 16. FIG. 7 is a diagram for explaining an evaluation method when performing comparative evaluation with a travel time storage table based on actual measured travel time data at the present time. First, referring to FIG. 6, a road section X for which a travel time is to be predicted and a time N of a prediction destination (time from the current time to a time to be predicted) N are input (S1).

Next, an LM-dimensional pattern vector is created based on the latest M actual measured travel time data up to the current time of the L road sections including the road section X (S2;
FIG. 7 (b)). The method of selecting the L road sections is arbitrary, but it is considered natural to select from the nearby road sections connected to the road section X. In FIG. 7B, the current time is 13: 3
The pattern vectors when 0, X = 1, L = 4, and M = 3 are shown. In addition, the time corresponding to the latest travel time data in the pattern vector (in the example of FIG.
3:30) is referred to as the “time of the pattern vector”.

The past travel time storage table stored in the variation pattern storage section 17 is read (S3). Here, various methods of selecting the travel time storage table can be considered. A running time storage table for several days in the past may be selected, a past running time storage table for the same day of the week as the current day may be selected, and a running time storage table for the same past weather or the same event may be selected. May be selected. A pattern vector including M pieces of travel time data of L road sections including the road section X is created. This pattern vector is created for a plurality of times in the travel time storage table. The method of selecting “a plurality of times” may be, for example, from a range of a fixed time before and after the current time, or may be all times in the travel time storage table.

This pattern vector is created in the same way for the travel time storage table for other days stored in the variation pattern storage section 17 (S4). An evaluation value (how similar) of the degree of approximation between the actually measured pattern vector obtained in S2 and the pattern vector in the travel time storage table is quantitatively obtained. This operation can be performed, for example, by obtaining a distance between vectors and using the distance as an evaluation value. A pattern vector having the highest evaluation value is specified in the travel time storage table (see FIG. 7A. The travel time storage table including the pattern vector is referred to as P).
The time T of the pattern vector is obtained (S5). In the travel time storage table P, the travel time Y of the road section X at the time obtained by adding the time N of the prediction destination to the time T obtained as described above is calculated as the travel time prediction value N time ahead of the current time. (S6).

As described above, it is possible to predict the travel time ahead of the current time by using the actually measured travel time up to the current time and the past travel time storage table. Next, a method of evaluating the reliability of the predicted traveling time will be described. FIG.
FIG. 9 and FIG. 9 are flowcharts for explaining the reliability evaluation method of the traveling time. First, a road section X for which a travel time is to be predicted, a prediction destination time (time from the current time to a time to be predicted) N, and a prediction error W are input (T
1).

Next, an LM-dimensional pattern vector is created based on the latest M actual measured travel time data up to the current time of the L road sections including the road section X (T
2). The past travel time storage table stored in the variation pattern storage unit 17 is read (T3), and the road section X is read.
A pattern vector of M pieces of travel time data of L road sections including is created. This pattern vector is created for a plurality of times in the travel time storage table. The creation of the pattern vector is similarly performed for the travel time storage table for other days stored in the variation pattern storage unit 17 (T4).

The measured pattern vector obtained at T2 and T
An evaluation value of the degree of approximation to the pattern vector in the travel time storage table obtained in step 4 is quantitatively obtained. In each of the travel time storage tables, K pattern vectors are specified in descending order of the evaluation value, and the travel time storage tables including those pattern vectors are stored in P (1),. . . , P (K). Then, the time T (1),. . . ,
T (K) is obtained (T5).

The time T obtained as described above
(1),. . . , T (K) plus the time N of the prediction destination, the travel time Y of the road section X
(1),. . . , Y (K) are obtained as the predicted travel time N hours ahead of the current time (T6). As a result, the K traveling times can be obtained in the descending order of the likelihood. Next, referring to FIG. 9, K traveling times Y (1),. . . , The travel time predicted value this seeking mode value Y ^* of Y (K) (T7). Y ^*
Is determined, for example, by the running times Y (1),. . . , Y (K), or a weighted average of each evaluation value.

Table 1 below shows the running times Y (1),. . . , Y
It is a table | surface which shows the specific example of (K). K = 5, Y (1) = 240, Y (2) = 200, Y (3) = 250, Y (4) = 24 in the descending order of the evaluation value.
0, Y (5) = 270 (unit is second). Y ^* = 240.

[0026]

[Table 1]

The running times Y (1),. . . , Y (K), and the probability falling within the range of Y ^* ± W (W is the input prediction error), the probability that the prediction error is within ± W can be obtained. The specific method is shown in FIG.
This will be described with reference to FIG. In FIG. 10, a specific example of Table 1 is used. W = 30 seconds. Y (1) ± 30, Y
(2) ± 30,. . . , Y (5) ± 30 are indicated by arrows in FIG.

In FIG. 10, when a ratio within the range of Y ^* ± 30 in the range indicated by each arrow is obtained (T8), this ratio can be set as the probability that the prediction error is within ± W ( T10). By setting W to a plurality of values, the prediction error
If the probabilities within W are obtained, and the distribution of these probabilities is used, the value of the prediction error falling within a certain probability can be obtained. For example, the probability can be determined value W ₉₀ of the prediction error entering within 90%. Generally speaking, it is possible to obtain a value WS of a prediction error whose probability is within _S.

[0029] based on the value W _S of the prediction error, it is possible to evaluate the travel time Y ^* of reliability predicted. The value W _S of the prediction error can be applied as follows. For example, when calculating the optimal route and the travel time required to travel on the route, the travel time Y ^{* of} each road is calculated ^.
If + W _S is calculated with S being large, the traveling time becomes longer, so that it is possible to reach the destination with sufficient time for traffic congestion.

Although the embodiment of the present invention has been described above, the embodiment of the present invention is not limited to the above embodiment. For example, in the above-described embodiment, the travel time data of L road sections including the road section X is compared.
The travel time data of only the road section X may be compared with L = 1. In addition, various changes can be made within the scope of the present invention.

[Brief description of the drawings]

FIG. 1 is a perspective view showing a road network including an intersection, and installation states of a beacon communication device 23, an ultrasonic vehicle sensor 24, and a camera type license plate detection device 25 on the road network.

FIG. 2 is a detailed diagram of a beacon communication device 23.

FIG. 3 is a configuration diagram of an entire system including a traveling time prediction device 13.

FIG. 4 is a functional block diagram of the traveling time prediction device 13.

FIG. 5 is a diagram showing a specific example of a travel time storage table.

FIG. 6 is a flowchart for explaining a travel time prediction method performed by an evaluation unit 15 and a travel time determination unit 16;

FIG. 7 is a diagram for explaining an evaluation method in the case of performing comparative evaluation with a travel time storage table based on currently measured travel time data.

FIG. 8 is a flowchart for explaining a reliability evaluation method of a traveling time.

FIG. 9 is a flowchart (continuation of FIG. 8) for explaining the reliability evaluation method of the traveling time.

FIG. 10 is a diagram for explaining a method of obtaining a probability that the prediction error is within ± W.

[Explanation of symbols]

 Reference Signs List 12 Data tabulation unit 13 Travel time prediction device 14 Storage unit 15 Evaluation unit 16 Travel time determination unit 17 Fluctuation pattern storage unit 18 Error calculation unit 23 Beacon communication device, 24 Ultrasonic vehicle sensor, 25 Camera type license plate detection device

Claims (5)

[Claims] 1. A running time prediction method comprising the following steps (a) to (d). (a) The road section X for which the travel time is to be predicted and the time N from the current time to the predicted time are determined. (b) One or more road sections including the road section X up to the current time. Collecting time data, (c) searching for a pattern similar to the travel time data and travel time collected by the procedure of (b) from the travel time storage tables collected in the past in the road section, (d) In the travel time storage table, the travel time storage value of the road section X N hours ahead of the pattern having a similar travel time is set as the travel time prediction value. 2. The travel time prediction method according to claim 1, wherein in (b), the range of time during which the travel time data is collected is from a time before a predetermined time in the past to the current time. 3. In (c), a plurality of similar pattern candidates are searched, and the procedure of (d) is executed for each of them. The travel time prediction method according to claim 1, wherein the method is determined as a travel time prediction value. 4. A running time prediction method comprising the following steps (e) and (f). (e) A travel time predicted value of the road section X whose travel time is to be predicted is determined based on past statistical travel time data. (f) A reliability of the travel time predicted value is determined. 5. A travel time prediction method comprising the following steps (g) to (k). (g) The road section X whose travel time is to be predicted and the time N from the current time to the predicted time are determined. (h) One or more road sections including the road section X up to the current time. Collecting time data, (i) searching a plurality of pattern candidates having similar running times to the running time data collected by the procedure of (h) from the running time storage tables collected in the past in the road section; (J) the travel time storage value of the road section X N hours ahead of a pattern having a similar travel time in the travel time storage table as a candidate travel time prediction value; (k) the travel time The mode value of the predicted value candidate value is determined as the travel time predicted value, and the reliability of the travel time predicted value is determined based on the spread of the distribution of the travel time predicted value candidate values.