WO2015059877A1

WO2015059877A1 - Information processing device and information processing method

Info

Publication number: WO2015059877A1
Application number: PCT/JP2014/005041
Authority: WO
Inventors: 智晴竹内
Original assignee: 三菱電機株式会社
Priority date: 2013-10-24
Filing date: 2014-10-03
Publication date: 2015-04-30
Also published as: JPWO2015059877A1; JP6107964B2

Abstract

An information processing device according to the present invention is provided with: a traffic data estimation unit which estimates, from probe data collected from vehicles traveling along links, time-series data for the link transit time of each link, where the term "link" refers to a road segment and the term "link transit time" refers to the time required to transit a link; and a traffic data interpolation unit which calculates the correlation coefficients between the link transit time of a link to be interpolated, the time-series data for which has missing data, and the link transit times of a plurality of links to be applied, the time-series data for which does not have any missing data and is used for interpolation, and then calculates interpolation data for interpolating data at missing time points in the time-series data for the link transit time of the link to be interpolated, on the basis of the calculated correlation coefficients and the link transit times of the plurality of links to be applied. This makes it possible to accurately interpolate data for links in an area, regardless of the proportion of links for which all data is not available.

Description

Information processing apparatus and information processing method

It relates to an information processing device that interpolates traffic information data.

The car navigation device provides the driver with the map information and the time it takes to reach the destination and the time it takes to pass through the traffic. Some smartphone applications have similar functions. These times are calculated based on the past or present traffic situation of the road section (hereinafter referred to as a link), the link passage time, and the like. For example, it is calculated based on traffic data transmitted by VICS (registered trademark) (Vehicle Information and Communication System). VICS (registered trademark) is a system that aggregates vehicle travel data collected from road sensors installed on a road and provides the data as traffic data. Since installation and maintenance of road sensors are expensive, installation of road sensors is currently limited to major trunk lines such as highways and national roads. Therefore, the traffic information regarding the road where the road sensor is not installed is not provided from VICS (registered trademark).

As another method of collecting vehicle travel data, there is a method of mounting sensors on the vehicle and collecting travel data of the vehicle. A vehicle equipped with a sensor is called a probe car, and traveling data of the probe car is called probe data. The probe data includes position coordinates and speed of the vehicle. When probe data is used, probe data in a range in which the probe car has traveled can be collected, and thus a wider range of travel data than VICS (registered trademark) can be collected.
On the other hand, there are few probe cars at present. In addition, since the position and time zone where the probe car travels are random, probe data may be lost depending on the road section or time zone. Therefore, interpolation of probe data for missing road sections and time zones is performed.

Principal component analysis of past probe data is performed for each link group, and a highly correlated base is selected from the calculated multiple bases, and linear synthesis is performed to estimate the interpolated data of the probe data of the link group including the missing link. Techniques to do this are disclosed. (See Patent Document 1 below)

Japanese Patent No. 4729469

However, in Patent Document 1, since the base is calculated for each area composed of the link group, the accuracy of the base deteriorates when the proportion of links that are biased in time increases although the total number of probe data is a certain amount or more. . A link with time bias is, for example, a case where a road is mostly used during morning and evening commuting hours. In this case, daytime and nighttime probe data is lost.

The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain an information processing apparatus that accurately interpolates data regardless of the proportion of links in which data is missing in an area. Yes.

A traffic data estimator that estimates time series data of link passage time indicating the time taken to pass the link from probe data collected from a vehicle traveling on a link that is a road section, and time series data of link passage time The correlation coefficient related to the link transit time between the interpolation target link that has a deficiency in the data and multiple interpolation application links that have no deficiency in the time series data of the link transit time is calculated. And a traffic data interpolation unit for calculating interpolation data for interpolating the link passing time at the missing time of the interpolation target link.

According to the present invention, data can be accurately interpolated regardless of the proportion of links in which data is missing in the area.

2 is a block diagram illustrating a configuration of an information processing device according to Embodiment 1. FIG. FIG. 5 shows an example of probe data according to the first embodiment. The figure which shows an example of the map data of map data DB which concerns on Embodiment 1. FIG. 3 is a flowchart showing a processing flow of the information processing apparatus according to the first embodiment. FIG. 4 is a diagram showing an example of time-series data of link passage times according to the first embodiment. FIG. 4 is a diagram showing an example of time-series data of link passage times according to the first embodiment. 4 is a flowchart showing a process flow of a traffic data estimation unit according to the first embodiment. The figure which shows the case where data are missing in the time series data of the link passage time according to the first embodiment and the first determination condition is satisfied. The figure which shows the case where data are missing in the time series data of the link passage time according to the first embodiment and the second determination condition is satisfied. The figure which shows the case where data are missing in the time series data of the link passage time according to the first embodiment. The figure which shows an example of the statistical data of statistical data DB which concerns on Embodiment 1. FIG. 5 is a flowchart showing a process flow of a traffic data interpolation unit according to the first embodiment. 9 is a flowchart showing a process flow of a traffic data estimation unit according to the second embodiment. The figure which shows the example which divided | segmented the time series data of the link passage time concerning Embodiment 2 into four time zones. 9 is a flowchart showing a process flow of a traffic data interpolation unit according to the second embodiment. The figure which shows the link to which the value of the correlation coefficient which concerns on Embodiment 2 is 2nd. The figure which shows the value of the correlation coefficient which concerns on Embodiment 2 to the 2nd place. The figure which shows the graph of the time series data of the link passage time until the value of the correlation coefficient which concerns on Embodiment 2 is 2nd. The figure which shows the example which matches the time series data of the link passage time divided | segmented into the four time zones which concern on Embodiment 2. FIG. FIG. 4 is a block diagram illustrating a configuration of an information processing device according to a third embodiment.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing the configuration of the information processing apparatus 1 according to the first embodiment. The information processing apparatus 1 includes a probe data receiving unit 10, a probe data processing unit 11, a map data database (hereinafter referred to as map data DB) 12 as a map data storage unit, a traffic data estimation unit 13, a traffic data output unit 14, It comprises a statistical data database (hereinafter referred to as statistical data DB) 15 as a statistical data storage unit and a traffic data interpolation unit 16.

Probe data is input to the probe data receiving unit 10 as vehicle travel data. The probe data is data collected from a plurality of probe cars. The probe data includes physical quantity data such as the position, travel speed, and traveling direction of the vehicle, and estimated data such as the type of road and the road surface condition. The physical quantity data is a value measured by a sensor device such as GPS (Global Positioning System) or a gyro sensor. The estimated data is data estimated from values measured by the sensor device.

The probe data receiving unit 10 outputs the received probe data to the probe data processing unit 11. The probe data processing unit 11 refers to the map information held in the map data DB 12 and determines on which link each probe data is acquired. The probe data processing unit 11 classifies the probe data for each link and outputs it to the traffic data estimation unit 13. The map data DB 12 holds map information indicating the link position.

The traffic data estimation unit 13 aggregates the classified probe data and estimates the time series data of the link transit time for each link. In many cases, the time-series data of the link transit time changes periodically with a period of 24 hours. Therefore, in the present embodiment, the traffic data estimation unit 13 estimates time-series data for 24 hours, which is one day, but may not be one day. The traffic data estimation unit 13 determines whether or not there is a deficiency in the time series data of the link passage time, and if there is no deficiency, outputs to the traffic data output unit 14. Data missing is a case where there is no link transit time value at a certain time. When there is a defect, the traffic data estimation unit 13 outputs the link as an interpolation target link to the traffic data interpolation unit 16. Moreover, the traffic data estimation part 13 outputs the time series data of the estimated link passage time to the statistical data DB 15 regardless of the presence or absence of data loss.

Statistic data DB 15 holds time-series data of the link passage time of each link as a statistic value. When new data is input for a link that already holds data, the statistical data DB 15 updates the statistical value. For example, the statistical data DB 15 obtains an average value with the value already held and holds it as a new statistical value. By holding it as a statistical value, the amount of data can be kept low. The traffic data interpolation unit 16 refers to the time series data of the link passage times of a plurality of links held in the statistical data DB 15 and interpolates the time series data of the link passage times of the interpolation target links. The traffic data interpolation unit 16 outputs the interpolated link passage time series data to the traffic data output unit 14.

The traffic data output unit 14 outputs time-series data of the link passage time for each link. The output data is used, for example, in a traffic data use service. The traffic data utilization service is a service for distributing traffic information to car navigation systems. The car navigation system calculates the time required to arrive at the destination based on the link passing time at the current time. Further, the output data is used for calculating a traffic jam degree indicating an average speed and a traffic jam degree when passing through the link. Furthermore, it may be used to determine the presence or absence of traffic. The traffic data output unit 14 may change the format and timing of data to be output according to the output destination.

FIG. 2 is a diagram illustrating an example of the probe data 21 according to the first embodiment. The probe data 21 includes a vehicle ID, longitude, latitude, traveling direction, road type, date and time, and travel speed. The vehicle ID is a value for identifying each vehicle. The longitude and latitude indicate the position of the vehicle. The traveling direction is a value indicating the direction in which the vehicle travels. The road type is a value indicating the type of road such as an expressway or a general national road. The date / time indicates the date / time when the data was acquired. The traveling speed is a value indicating the speed of the vehicle. The longitude, latitude, traveling direction, and road type are map information. The date / time and the traveling speed are traveling information.

Probe data 21 a to e are specific examples of data stored as probe data 21. For example, in the probe data 21a, a vehicle having a vehicle ID of 0001 is traveling on a highway in a direction of 10 ° at a position of longitude 35.000 and latitude 139.000, and the date and time is 9 o'clock on September 1, 2013. The traveling speed is 90 km / hour.

Probe data 21 is an example. The probe data only needs to include at least map information that identifies the link on which the vehicle is traveling and traveling information that can estimate the time required for the vehicle to pass through the link. In addition, if the item of the map information of the probe data 21 includes the item of the map data DB 12, the link on which the vehicle is traveling can be easily specified. For example, when the map data DB 12 is provided with latitude and longitude as items, if the latitude and longitude are included as the map information of the probe data 21, the link on which the vehicle is traveling can be easily specified.

The map data DB 12 is a database that stores link map data. The link map data is data indicating the position of the link. The map data includes at least a link number that identifies the link, and a latitude and longitude that indicate the position of the link. The map data may include traffic restriction information such as a speed limit and a traveling direction, a road type that distinguishes an expressway, a general national road, a prefectural road, and the like. By adding the traffic restriction information and the road type to the map data, the accuracy of associating the links with the probe data 21a-n is improved.

FIG. 3 is a diagram illustrating an example of the map data 31 of the map data DB 12 according to the first embodiment. The map data 31 includes a forward link number, a reverse link number, a road number, longitude, latitude, and a road type. The forward link number is a value indicating the uplink lane of the link. The reverse link number is a value indicating the downlink lane of the link. The road number is a value for identifying the road. The longitude is a value indicating the longitude of the link. The latitude is a value indicating the latitude of the link. The road type is a value indicating the type of road such as an expressway or a general national road.

The map data 31 a to f are specific examples of data stored as the map data 31. For example, the map data 31a indicates that the forward link number is 100, the reverse link number is 103, the road number is 10, the longitude is 35.000, the latitude is 139.200, and the road type is a general national road. .
The map data 31a to 31c are data relating to the same road having the same road number 10. Although it is the data regarding the same road, since the position where data was collected is different, different values for longitude and latitude are stored.

Next, the operation of the information processing apparatus 1 will be described.
FIG. 4 is a flowchart showing a process flow of the information processing apparatus 1 according to the first embodiment. When receiving the probe data, the information processing apparatus 1 starts processing from step S41.

In step S41, the probe data receiving unit 10 receives the probe data 21. The timing at which the probe data 21 is input to the probe data receiving unit 10 is not particularly limited. The probe data 21 may be input regularly or irregularly. Further, the smaller the measurement interval of the probe data 21 is, the higher the accuracy of the link passing time can be estimated. If the measurement interval is small, the amount of data increases, and fluctuations in the estimated value of the link transit time can be reduced. The measurement interval is arbitrary, but it is desirable to set it at several seconds. The probe data receiving unit 10 outputs the probe data 21 to the probe data processing unit 11. The process proceeds to step S42.

In step S42, the probe data processing unit 11 classifies the input probe data 21a to 21n for each link. The association between the link and each probe data 21a-n is called map matching. The probe data processing unit 11 determines and classifies which link each probe data 21a to 21n is running on the basis of the map data DB 12 and the map information of each probe data 21a to 21n. Further, the probe data processing unit 11 calculates the length of each link. The length of each link is a road. The probe data processing unit 11 outputs the probe data 21a to 21n classified for each link and the length of each link to the traffic data estimation unit 13. The process proceeds to step S43.

In step S43, the traffic data estimation unit 13 aggregates the probe data classified for each link, and estimates the time series data of the link passage time for each link. The link passing time is the time taken for the link to pass. The time series data of the link passage time represents the daily time series change with respect to the link passage time. Details of the processing will be described later. The process proceeds to step S44.

In step S44, the traffic data estimation unit 13 determines whether or not interpolation is required for the time series data of the link passage time for each link.
FIG. 5 is a diagram illustrating an example of time-series data of the link passage time according to the first embodiment. The horizontal axis is time. The vertical axis is the link passing time. This is an example in which there is no missing data. Since there is no deficiency in the data, the data exists at equal intervals.
The traffic data estimation unit 13 determines that interpolation is not necessary when data is not missing. The process proceeds to step S46 in FIG.

FIG. 6 is a diagram illustrating an example of time-series data of link passage times according to the first embodiment. The horizontal axis is time. The vertical axis is the link passing time. This is an example of missing data. Since there is a deficiency in data, there are places where data does not exist at regular intervals.
If the data is missing, the traffic data estimation unit 13 determines that interpolation is necessary, and sets the link as an interpolation target link. The process proceeds to step S45 in FIG.

4, the traffic data interpolation unit 16 interpolates the time series data of the link passage time with respect to the interpolation target link. Details of the processing will be described later. The process proceeds to step S46.

In step S46, the traffic data output unit 14 outputs time series data of the link passage time for each link. The process ends.

Next, the operation of the traffic data estimation unit 13 will be described.
FIG. 7 is a flowchart showing a process flow of the traffic data estimation unit 13 according to the first embodiment. It is the detail of step S43 of FIG. The traffic data estimation unit 13 starts processing from step S71 when the probe data 21a to n classified for each link and the length of each link are input from the probe data processing unit 11.

In step S71, the traffic data estimation unit 13 aggregates the probe data 21a to 21n classified for each link and estimates time series data of the link passage time. The traffic data estimation unit 13 calculates an average passing speed at each time for each link based on the date and traveling speed of the probe data 21a to 21n. The traffic data estimation unit 13 estimates the time series data of the link passage time by calculating the average passage time for each time from the length of each link. At this time, if the sampling interval is set to be as small as about 5 minutes, it is possible to confirm the change in the traffic congestion state of the link. The traffic data estimation unit 13 may estimate the time series data of the link transit time separately according to weekdays, holidays, days of the week, and the like.

In addition, fluctuations such as noise may occur when the time series data of the link transit time is estimated. This is because, as a general characteristic of a traveling vehicle, even if the vehicle travels on the same link at the same time, the actual traveling speed differs from one traveling vehicle to another, and thus data varies. In order to eliminate this fluctuation, the traffic data estimation unit 13 may smooth the time series data of the link passage time. For example, the traffic data estimation unit 13 smoothes the data by averaging the link passage time data for 30 minutes before and after each sampled time. However, if the averaging time width is too large, link characteristics such as traffic congestion will be lost. Therefore, it is necessary to perform averaging by setting an appropriate time interval from the estimation accuracy and processing cost.
The process proceeds to step S72.

In step S72, the traffic data estimation unit 13 determines the presence or absence of data loss in the time series data of the link passage time. When there is a data loss, the traffic data estimation unit 13 determines a data loss rate. The time at which data is missing is called the missing time, and the rate at which the data is missing is called the degree of deficiency. Two conditions for determining the height of the deficiency are illustrated.

First, the first determination condition will be described. The traffic data estimation unit 13 determines that the deficiency is high when the total time during which data is missing is greater than or equal to a threshold value in a certain time width.

FIG. 8 is a diagram illustrating a case where data is missing in the time series data of the link passage time according to the first embodiment and the first determination condition is satisfied. FIG. 8 shows an example in which the sampling interval of the link passage time is 5 minutes, the time width for loss determination is 30 minutes, and the threshold is 20 minutes. Since the sampling interval is 5 minutes, 6 points of data should exist in a time width of 30 minutes. However, in the time width of 30 minutes indicated by the dotted frame in FIG. 8, there are only two points of data, and a total of 20 minutes of data is missing. Since the threshold value is 20 minutes, the traffic data estimation unit 13 determines that the degree of data loss is high in the time span.

Note that the time span for determining the deficiency may be the entire time-series data of the link passage time.
Further, the traffic data estimation unit 13 shifts the time width little by little for the entire time series data of the link passage time, and determines the data loss degree. When the number of times that the loss degree is determined to be high is equal to or greater than the threshold, the link It may be determined that the degree of deficiency of the entire time series data of the passage time is high.

Next, the second determination condition will be described. The traffic data estimation unit 13 determines that the deficiency is high when the total time for which data is continuously missing is equal to or greater than a threshold.

FIG. 9 is a diagram illustrating a case where data is missing in the time series data of the link passage time according to the first embodiment and the second determination condition is satisfied. FIG. 9 shows a case where, for example, the sampling interval of the link passage time is 5 minutes and the threshold of the continuous missing time is 15 minutes. Since the sampling interval is 5 minutes, it is determined that the degree of data loss is high when three or more points of data are continuously lost. At 15 minutes indicated by the arrow in FIG. 9, data is missing. The traffic data estimation unit 13 determines that the time series data has a high degree of deficiency.

In the present embodiment, the traffic data estimation unit 13 determines that the degree of data loss is high when one of the two determination conditions is satisfied, but may determine the data by other methods.

FIG. 10 is a diagram showing a case where data is not lost in the time series data of the link passage time according to the first embodiment. In FIG. 10, although 10 minutes of data is missing twice, neither the first determination condition nor the second determination condition is satisfied. Therefore, the traffic data estimation unit 13 determines that the data deficiency is low.

The traffic data estimation unit 13 sets the link as an interpolation target link when the degree of data loss is low. The interpolation when the data deficiency is high is described in the second embodiment. The process of FIG. 7 proceeds to step S73.

7, the traffic data estimation unit 13 outputs time-series data of the estimated link passage time to the statistical data DB 15. The process proceeds to step S74.

In step S74, the traffic data estimation unit 13 outputs the time series data of the link passage time of the interpolation target link to the traffic data interpolation unit 16. The traffic data estimation unit 13 outputs the time series data of the link passage times of links that are not interpolation targets to the traffic data output unit 14. The process ends.

FIG. 11 is a diagram illustrating an example of statistical data of the statistical data DB 15 according to the first embodiment. The statistical data DB 15 holds time-series data of link transit times input from the traffic data estimation unit 13. In each graph 111a-n in FIG. 11, the horizontal axis represents time, and the vertical axis represents link passage time.

Next, the operation of the traffic data interpolation unit 16 will be described.
FIG. 12 is a flowchart showing a processing flow of the traffic data interpolation unit 16 according to the first embodiment. The traffic data interpolation unit 16 interpolates the time series data of the link transit time of the link determined as the interpolation target link by the traffic data estimation unit 13. The traffic data interpolation unit 16 starts processing from step S121 when time series data of the link passage time of the interpolation target link is input.

In step S121, the traffic data interpolation unit 16 refers to the statistical data DB 15 and extracts time-series data of the link transit times of links for which no data is missing. The extracted link is called a reference link. The traffic data interpolation unit 16 extracts the reference link without considering the positional relationship between the reference link and the interpolation target link. The area may be arbitrarily determined. For example, you may divide | segment into a mesh shape on a map. The size of the area may be changed according to the number of links in which data is not missing.

For example, areas where there is a lot of traffic and where probe data is likely to gather, such as in front of a station or near a main road, may be reduced, while areas such as rural areas and mountainous areas where traffic is low may be enlarged. By changing the size of the area, the traffic data interpolation unit 16 can secure a certain number of reference links. The traffic data interpolation unit 16 may use all links without data loss among the links stored in the statistical data DB 15 as reference links, but the processing cost increases. Therefore, the traffic data interpolation unit 16 may extract the reference link from the area to which the interpolation target link belongs. The process proceeds to step S122.

In step S122, the traffic data interpolation unit 16 calculates the correlation coefficient between the time series data of the link passage time of the interpolation target link and the time series data of the link passage time of the reference link. If the data deficiency is low, the correlation coefficient is calculated by thinning out the data, or for example, the traffic data interpolation unit 16 is deficient in the intermediate value of the data at the time before and after the deficient data. The correlation coefficient is calculated as follows. A correlation coefficient is calculated for each of the plurality of reference links. Further, the traffic data interpolation unit 16 ranks the reference links in descending order of the correlation coefficient, and extracts data up to the R rank. The process proceeds to step S123.

In step S123, the traffic data interpolation unit 16 standardizes the time series data of the link passage times of the reference links up to the R rank. Since each link has a different length, the link transit time differs when passing through each link at the same speed. Therefore, for example, the time series data of the link passage time is standardized by a value obtained by dividing the link length by the regulation speed. The traffic data interpolation unit 16 can match the length of the link passage time of the interpolation target link by multiplying the normalized time series data of the link passage time by the length of the interpolation target link. The process proceeds to step S124.

In step S124, the traffic data interpolation unit 16 sets the reference link having the correlation coefficient value up to the R rank as the interpolation application link. The traffic data interpolation unit 16 calculates the interpolation data by adding the weighted time series data of the link transit times of the plurality of interpolation application links. The traffic data interpolation unit 16 calculates the interpolation data L0 (T) using Equation 1. L (T, r) indicates time-series data of the standardized link transit time. x (T, r) represents a correlation coefficient.

The traffic data interpolation unit 16 substitutes the interpolation data as the missing time value. In addition, the traffic data interpolation unit 16 smoothes data in an arbitrary time width in order to maintain continuity. The process proceeds to step S125.

In step S125, the traffic data interpolation unit 16 returns to step S121 if there are other interpolation target links. If there is no other interpolation target link, the traffic data interpolation unit 16 ends the process.

In the present embodiment, the statistical data DB 15 stores the statistical values for the time series data of the link transit time of each link. However, the statistical data DB 15 may hold the data or the statistical values by dividing the period. . For example, by using data of the past several hours, it is possible to estimate the link transit time according to the current situation such as an accident or traffic jam. Further, by using data when a similar event in the past is used as the link passage time when an event such as a fireworks display or a concert is held, it is possible to estimate the link passage time with higher accuracy.

Therefore, in the present embodiment, a traffic data estimation unit that estimates time series data of a link passing time indicating a time taken to pass the link from probe data collected from a vehicle traveling on a link that is a road section; , Calculate the correlation coefficient for the link transit time between the link to be interpolated that has a deficiency in the time series data of the link transit time and a plurality of interpolation application links that have no deficiency in the time series data of the link transit time. A traffic data interpolating unit that calculates interpolation data for interpolating the link passing time of the missing time of the interpolation target link based on the link passing time and the correlation coefficient, so that data is missing in the area Regardless of the link ratio, the accuracy of data interpolation can be increased.

Also, interpolation data is calculated from time-series data of link passage times of links with higher correlation by making the top N correlation coefficients (N is an integer equal to or greater than 1) as an interpolation application link. The accuracy in data interpolation can be increased.

In addition, by calculating the interpolation data from a plurality of interpolation application links, it becomes difficult to reflect the characteristic unique to the time series data of the link transit time of the interpolation application link in the interpolation data.

In addition, the traffic data estimation unit divides a map including a plurality of links into a plurality of regions, and determines an interpolation application link from links included in the same region as the interpolation target link. Processing costs can be reduced.

In addition, since the traffic data estimation unit determines the size of the area according to the amount of probe data collected in the area, a certain amount of probe data can be secured. Therefore, it is possible to improve accuracy in data interpolation.

Embodiment 2. FIG.
In the first embodiment described above, the time series data of the link transit time is interpolated using data of another link having a high correlation when the data is missing. Shows an embodiment in which time-series data of link transit times is divided into a plurality of time zones, and interpolation is performed using link data having a high correlation in other time zones when the rate of data loss is high. .
In the present embodiment, all the configurations described in the first embodiment as shown in FIG. 1 are provided, and further additional configurations will be described.
Since the configuration of the information processing apparatus 1 is the same as that of the first embodiment, the description thereof is omitted.

The operation of the traffic data estimation unit 13 will be described.
FIG. 13 is a flowchart showing a process flow of the traffic data estimation unit 13 according to the second embodiment. When the probe data 21a-n classified for each link and the length of each link are input from the probe data processing unit 11, the traffic data estimation unit 13 starts processing from step S131.

Step S131 is the same as step S71 in FIG. The process proceeds to step S132.

In step S132, the traffic data estimation unit 13 divides the time series data of the link passage time into a plurality of time zones. The number of divisions is not particularly limited. The process proceeds to step S133.

In step S133, the traffic data estimation unit 13 determines the data loss degree in each of a plurality of time zones. The determination of the degree of data loss is the same as the method described in step S72 in FIG.
In the present embodiment, even when the degree of data loss is high, the interpolation is performed. Details will be described later. The process proceeds to step S134.

The processes in steps S134 to S135 are the same as those in steps S73 to S74 in FIG. 7 for the time series data in each of a plurality of time zones. Description is omitted. The process ends.

FIG. 14 is a diagram illustrating an example in which the time-series data of the link passage time according to the second embodiment is divided into four time zones. 141a is time series data of the link passage time for one day of the interpolation target link. 141b is time-series data of the link transit time in the first time zone T1. There is no missing data and it is not an interpolation target link. 141c is time-series data of the link transit time in the second time zone T2. Since the degree of data loss is high, the correlation with the reference link cannot be calculated. 141d is time-series data of the link transit time in the third time zone T3. There is no missing data and it is not an interpolation target link. 141e is time-series data of the link transit time in the fourth time zone T4. Since the data deficiency is low, the interpolation data can be calculated by calculating the correlation with the reference link.

It should be noted that, if the degree of deficiency is high in any time zone and the correlation calculation with the reference link cannot be performed in all time zones, it is not a target of this embodiment.

Next, the operation of the traffic data interpolation unit 16 will be described. The case shown in FIG. 14 will be described as an example.

FIG. 15 is a flowchart showing a processing flow of the traffic data interpolation unit 16 according to the second embodiment. The traffic data interpolation unit 16 interpolates the time series data of the link transit time of the link determined as the interpolation target link by the traffic data estimation unit 13. The traffic data interpolation unit 16 starts processing from step S151 when time-series data of the link passage time of each time zone including the interpolation target link is input.

Since step S151 is the same as step S121 of FIG. 12, description thereof is omitted. The process proceeds to step S152.

In step S152, the traffic data interpolation unit 16 calculates a correlation with the reference link for a time zone in which there is no data loss and a time zone in which the data loss degree is low. A correlation coefficient between the time series data of the link passage times of the time zones T1, T3, and T4 and the time series data of the link passage times of the same time zone of the reference link is calculated. The time zone T4 is a case where the degree of data loss is low. In this case, the traffic data interpolation unit 16 calculates a correlation coefficient as data that lacks an intermediate value of data at times before and after the missing data. In the time zone T2, since data loss is high, the correlation cannot be calculated.

The traffic data interpolation unit 16 ranks the reference links in descending order of the correlation coefficient, and extracts data up to the R rank. As an example, graphs of links up to the second place, correlation coefficient values, and time-series data are shown in FIGS.

FIG. 16 is a diagram showing links up to the second place of correlation coefficient values according to the second embodiment. In the time zone T1, the first place is the link L1, and the second place is the link L2. The time zone T2 has no value because the correlation cannot be calculated. In the time zone T3, the first place is the link L1, and the second place is the link L3. In the time zone T4, the first place is the link L2, and the second place is the link L3.

FIG. 17 is a diagram showing values up to the second place of the correlation coefficient according to the second embodiment. In the time zone T1, the first-order correlation coefficient is a1, and the second-order correlation coefficient is a2. The time zone T2 has no value because the correlation cannot be calculated. In the time zone T3, the first-order correlation coefficient is c1, and the second-order correlation coefficient is c2. In the time zone T4, the first-order correlation coefficient is d1, and the second-order correlation coefficient is d2.

FIG. 18 is a diagram showing a graph of time-series data of the link transit time up to the second-order correlation coefficient value according to the second embodiment. The first place in the time zone T1 is 181a, and the second place is 181b. The first place in the time zone T3 is 181c, and the second place is 181d. The first place in the time zone T4 is 181e, and the second place is 181f.

In FIG. 15, the process proceeds to step S153.
In step S153, the traffic data interpolation unit 16 normalizes the time series data of the link passage time of each link. The process is the same as step S123 in FIG. The process proceeds to step S154.

In step S154, the traffic data interpolation unit 16 calculates the interpolation data of the time zone in which the data is missing by adding the weighted time series data of the link passing times of the plurality of interpolation application links. When the deficiency of the data in the time zone Tk is low, the traffic data interpolation unit 16 calculates the interpolation data L0 (Tk) using Equation 2. The traffic data interpolation unit 16 calculates the time series data L (Tk, r) of the link transit time until the correlation coefficient rank r reaches R, and the correlation coefficient x (Tk, r).

In the time zone T4, the traffic data interpolation unit 16 calculates the interpolation data L0 (T4) using Equation 3.

When the degree of data loss in the time zone Tk is high, the traffic data interpolation unit 16 calculates the interpolation data L0 (Tk) using Equation 4. The traffic data interpolation unit 16 uses the time series data L (Tk, r) of the link transit time up to the R rank of the correlation coefficient in the time zone other than Tk, and the correlation coefficient x (Tk, r). To calculate.

In the time zone T2, the traffic data interpolation unit 16 calculates the interpolation data L0 (T2) using Equation 5.

The traffic data interpolation unit 16 substitutes the interpolation data as the missing time value. In addition, the traffic data interpolation unit 16 smoothes data in an arbitrary time width in order to maintain continuity. The process proceeds to step S155.

In step S155, the traffic data interpolation unit 16 combines the interpolated time zone data and the non-interpolated time zone data to generate one day of interpolated data. At this time, the time series data in the interpolated time zone may lose continuity with the time series data in the preceding and following time zones. Therefore, the traffic data interpolation unit 16 smoothes the data before and after the interpolated time zone. For example, the traffic data interpolation unit 16 averages and smoothes the data for 30 minutes before and after each of the data included in the first 30 minutes and the last 30 minutes of the time zone.

FIG. 19 is a diagram illustrating an example in which time-series data of link passage times divided into four time zones according to the second embodiment are combined. 19a is time-series data of the link transit time in the first time zone T1. There is no missing data and no interpolation. 19b is time-series data of the link transit time in the second time zone T2. The data is missing and the data is interpolated. 19c is time series data of the link passing time in the third time zone T3. There is no missing data and no interpolation. 19d is the time series data of the link passing time in the fourth time zone T4. The data is missing and the data is interpolated. Reference numeral 19e denotes data for one day including the time series data of the link passage times of 19a to 19d.

In FIG. 15, the process proceeds to step S156.
In step S156, the traffic data interpolation unit 16 returns to step S151 if there are other interpolation target links. If there is no other interpolation target link, the traffic data interpolation unit 16 ends the process.

If the data deficiency is high, it is highly possible that the time-series data cannot express the characteristics of changes in the link transit time even if the intermediate value of the previous and subsequent data is taken. Therefore, even if the correlation with the reference link is calculated, an appropriate interpolation application link cannot be extracted. Therefore, it is more likely that the link passage time can be estimated with higher accuracy when the interpolation data is calculated using data in other time zones. Note that even when the data deficiency is low, interpolation data may be calculated using data in other time zones.

Therefore, in the present embodiment, the traffic data interpolation unit divides the time series data of the link passage time into a plurality of time zones, calculates a correlation coefficient regarding the link passage time with the interpolation application link for each time zone, Interpolation data is calculated based on the correlation coefficient in the time zone including the missing time and the link transit time of the interpolation application link, so that the interpolation application link with higher correlation can be used depending on the time zone, and in data interpolation Increase accuracy.

In addition, the traffic data interpolation unit divides the time series data of the link passage time into a plurality of time zones, calculates the correlation coefficient regarding the link passage time with the interpolation application link for each time zone, and does not include the missing time. Since the interpolation data is calculated based on the correlation coefficient in the band and the link passage time of the interpolation application link, the interpolation data can be calculated for more links. Therefore, the information processing apparatus 1 can output highly accurate traffic data for more links.

Moreover, since the traffic data estimation unit estimates the time series data of the link transit time of the interpolation application link based on the probe data of the first period, the current situation or the situation that occurred in the past specific period It is possible to estimate the link transit time according to

Embodiment 3 FIG.
In the second embodiment described above, the time series data of the link transit time is divided into a plurality of time zones, and when the data loss rate is high, interpolation is performed using data of links having high correlation in other time zones. In this embodiment, an embodiment in which data collected from road sensors is added to the data of the statistical data DB is shown.
In the present embodiment, all the configurations described in the first embodiment as shown in FIG. 1 are provided, and further additional configurations will be described.

FIG. 20 is a block diagram showing a configuration of the information processing apparatus 200 according to the third embodiment. The information processing apparatus 200 includes a probe data reception unit 10, a probe data processing unit 11, a map data DB 12, a traffic data estimation unit 13, a traffic data output unit 14, a statistical data DB 15, a traffic data interpolation unit 16, and a road sensor data reception unit 201. The road sensor data processing unit 202 is configured.

The road sensor data collected from the road sensor is input to the road sensor data receiving unit 201 as travel data. The road sensor data is based on data collected by road sensors such as vehicle detectors installed on the road. The road sensor observes a passing vehicle and collects data. The timing at which road sensor data is input is not particularly limited. The road sensor data may be input regularly or irregularly.

A typical example of road sensor data is VICS (registered trademark). In VICS (registered trademark), the link currency time estimated from the data collected by the road sensor every 5 minutes is transmitted to the center. The road sensor data receiving unit 18 outputs road sensor data to the road sensor data processing unit 202.

The road sensor data processing unit 19 aggregates the input road sensor data and outputs time-series data of the link passage time for each link to the statistical data DB 15.

The statistical data DB 15 holds the time series data of the link passage times of the links input from the traffic data estimation unit 13 and the road sensor data processing unit 202 as statistical values. When new data is input for a link that already holds data, the statistical data DB 15 updates the statistical value.

Therefore, in this embodiment, when road sensor data that is travel data of a vehicle traveling on a link collected by a road sensor installed on the link is input, it is classified for each link and the link passing time is calculated. It is equipped with a road sensor data processing unit that estimates time-series data, and a statistical data storage unit holds time-series data of link transit times input from the traffic data estimation unit and road sensor data processing unit as statistical values for each link. Therefore, since more time series data of link passage times can be used as a reference link, accuracy can be improved in interpolation of time series data of link passage times.

1,200 Traffic Information Processing Device 10 Probe Data Receiving Unit 11 Probe Data Processing Unit 12 Map Data DB
13 Traffic data estimation unit 14 Traffic data output unit 15 Statistical data DB
16 Traffic

data interpolation unit

21, 21a-

n Probe data

31, 31a-n Map data 111a-n, 141a-e, 181a-f, 191a-e Time series data of link passing time 201 Road sensor data reception unit 202 Road sensor Data processing section

Claims

A traffic data estimating unit that estimates time series data of a link passing time indicating a time taken to pass the link from probe data collected from a vehicle traveling on a link that is a road section;
Calculating a correlation coefficient relating to a link passage time between the interpolation target link having a deficiency in the time series data of the link passage time and a plurality of the interpolation application links having no deficiency in the time series data of the link passage time, and the interpolation A traffic data interpolation unit that calculates interpolation data for interpolating the link passage time of the missing time of the interpolation target link based on the link passage time of the application link and the correlation coefficient;
An information processing apparatus comprising:
The traffic data interpolation unit divides the time series data of the link passage time into a plurality of time zones, calculates the correlation coefficient regarding the link passage time with the interpolation application link for each time zone, and includes the missing time The information processing apparatus according to claim 1, wherein the interpolation data is calculated based on the correlation coefficient in a time zone and a link passage time of the interpolation application link.
The traffic data interpolation unit divides the time series data of the link passage time into a plurality of time zones, calculates the correlation coefficient regarding the link passage time with the interpolation application link for each time zone, and includes the missing time The information processing apparatus according to claim 1, wherein the interpolation data is calculated based on the correlation coefficient in a non-period of time and a link passage time of the interpolation application link.
4. The traffic data interpolation unit according to claim 1, wherein the link having the highest N correlation coefficients (N is an integer of 1 or more) is set as the interpolation application link. 5. The information processing apparatus described in 1.
The traffic data estimation unit is a deficiency indicating a rate of data deficiency when the total time for which data is missing in the first time width of the time series data of link passing time is equal to or greater than a first threshold. The information processing apparatus according to claim 1, wherein the information processing apparatus determines that the degree is high.
The traffic data estimation unit determines that the deficiency is high when data of a time equal to or greater than a second threshold is continuously missing in the time series data of the link transit time. To 5. The information processing apparatus according to any one of 5 to 5.
The traffic data interpolation unit, when the loss degree of the time series data of the link passage time of the divided time zone is high, the correlation coefficient in the time zone not including the loss time and the link passage time of the interpolation application link The information processing apparatus according to claim 3, wherein the interpolation data is calculated based on the information.
The traffic data estimation unit divides a map including a plurality of links into a plurality of regions, and determines the interpolation application link from links included in the same region as the interpolation target link. The information processing apparatus according to any one of claims 7 to 9.
The information processing according to any one of claims 1 to 8, wherein the traffic data estimation unit determines the size of the region according to an amount of the probe data collected in the region. apparatus.
The traffic data interpolation unit multiplies the correlation coefficient by the link transit time at the missing time obtained by correcting the length of the interpolation application link by the length of the interpolation target link, thereby adding the interpolation data. The information processing apparatus according to any one of claims 1 to 9, wherein the information processing apparatus is calculated.
A probe data processing unit that classifies the probe data for each link based on map information for identifying a link, and outputs the classified probe data to the traffic data estimation unit;
A statistical data storage unit that holds, as a statistical value, time-series data of link transit times input from the traffic data estimation unit;
The information processing apparatus according to claim 1, further comprising:
When road sensor data, which is travel data of a vehicle traveling on a link, collected by a road sensor installed on the link is input, the road sensor classifies each link and estimates time-series data of the link transit time. A data processing unit,
12. The statistical data storage unit holds time-series data of link passage times input from the traffic data estimation unit and the road sensor data processing unit as a statistical value for each link. The information processing apparatus according to any one of the above.
The said traffic data estimation part estimates the time series data of the link passage time of the said interpolation application link based on the said probe data of a 1st period, It is any one of Claim 1 to 12 characterized by the above-mentioned. The information processing apparatus described.
A traffic data estimation step for estimating time series data of a link passing time indicating a time taken to pass the link from probe data collected from a vehicle traveling on a link which is a road section;
Calculating a correlation coefficient relating to a link passage time between the interpolation target link having a deficiency in the time series data of the link passage time and a plurality of the interpolation application links having no deficiency in the time series data of the link passage time, and the interpolation A traffic data interpolation step for calculating interpolation data for interpolating the link passage time of the missing time of the interpolation target link based on the link passage time of the application link and the correlation coefficient;
An information processing method comprising: