JP4773823B2 - Traffic situation prediction method, apparatus and program thereof - Google Patents

Description

  The present invention relates to a traffic situation prediction method, an apparatus thereof, and a program for predicting a future traffic situation with high accuracy based on a current traffic situation transmitted from a traffic information center.

In recent years, with the spread of VICS (Vehicle Information and Communication System) that provides information on the current state of traffic, in-vehicle navigation devices that perform the shortest time route search using the current state information have become widespread.
However, in the current status information, when the traffic situation changes greatly with the passage of time, the accuracy is significantly deteriorated. Therefore, it is desired to predict the future traffic situation with high accuracy.

As an example of an existing technique for predicting a future traffic situation, there is a technique for predicting the future with reference to past traffic situation data and current situation information, for example. According to this technology, a past day that is close to the traffic situation is identified based on the similarity between the past date and the predicted date, and the traffic condition data of the specified past date at the prediction target time is used as a predicted value (for example, Patent Document 1). reference).
In addition, multiple traffic congestion prediction methods are used to select and output the traffic congestion status based on the prediction method with high prediction accuracy, regardless of the type or characteristics of the road such as expressways and general roads, or the length of the prediction destination time. In addition, a traffic jam prediction method that can predict the future traffic jam situation is also known (for example, see Patent Document 2).

Furthermore, using multiple different prediction models such as short-term prediction model, medium-term prediction model, long-term prediction, etc., the final prediction is performed by weighting the prediction values of each prediction means so that the weight increases as the prediction accuracy increases. There is also known a traffic situation prediction apparatus that can calculate a value and predict with high accuracy (see, for example, Patent Document 3).
Japanese Patent Laying-Open No. 2005-63034 (paragraph “0018” to “0037”, FIG. 1) Japanese Patent Laying-Open No. 2005-141661 (paragraphs “0009” to “0010”, FIG. 1) Japanese Patent Laying-Open No. 2005-227972 (paragraphs “0009” to “0031”, FIG. 1)

However, according to the technique disclosed in Patent Document 1, even if the traffic situation is close to the specific date in the past in a certain period, the traffic condition is not always close to the same specific date until the future. The expressway is greatly affected by sudden events such as traffic accidents, traffic regulations, road construction, and weather, and good accuracy cannot be expected in such situations.
Further, according to the technology disclosed in Patent Document 2, it is possible to predict the future traffic congestion situation with high accuracy regardless of the type and characteristics of the road such as a highway and a general road. Considering the case of searching for the route with the shortest travel time by predicting changes, if the prediction method is switched halfway, the prediction value obtained there will be discontinuous, resulting in a decrease in the accuracy of the route search. It has the inconvenience.

  Furthermore, according to the technique disclosed in Patent Document 3, since the final prediction value is calculated by combining the prediction values of a plurality of prediction models by weight addition, the prediction value due to the switching of the prediction model is abrupt. No change occurs. However, the traffic situation prediction is calculated from the correlation between the traffic situation in the prediction target section (link) and the traffic situation in the related link previously associated with this prediction target section. Is desired.

  The present invention has been made based on the various circumstances described above, and selects an appropriate prediction model according to the road type, and in particular, on highways, continuously with high accuracy regardless of the presence or absence of sudden events. It is an object of the present invention to provide a traffic situation prediction method, an apparatus for the same, and a program that can perform traffic situation prediction.

  In order to solve the above-described problem, for example, in the case of an expressway, the present invention provides a step in which a traffic situation prediction device captures current and past traffic information from the outside via a communication device, and for each prediction target link. One or more related propagation source links are associated, and for each propagation source link, a prediction parameter that constitutes a prediction model using a propagation phenomenon of a traffic situation composed of a propagation time and a weight coefficient is generated and stored in a storage device. Predicting the future traffic situation by a computing device based on the prediction model using the traffic information and the prediction parameter read from the storage device, and outputting the predicted traffic information via the communication device And step.

In addition, the map data including link attribute information is searched for a link that is connected to the prediction target link n-th order, and among the searched links, a link within a predetermined distance from the prediction target link, the link length is a predetermined length A step of setting a link corresponding to at least one of a longer link, a link whose link type is a main line, and a link for which traffic information is provided as the propagation source link.
The link order n in the n-order connection is defined as a link directly connected to the upstream side of the prediction target link as a primary link and a link directly connected to the upstream side of the primary link as a secondary link. This is the order when the link is directly connected to the upstream side of the secondary link as a tertiary link (n ≧ 1).

  According to the present invention, it is possible to select an appropriate prediction model according to the road type, and in particular, it is possible to predict traffic conditions continuously and with high accuracy regardless of the presence or absence of sudden events on expressways.

FIG. 1 is a diagram showing a schematic configuration of a traffic information system including a traffic situation prediction apparatus according to an embodiment of the present invention.
As shown in FIG. 1, the traffic information system includes an in-vehicle device 13 mounted on a vehicle 12, a computer 14 managed and operated by a traffic information center, and a traffic situation prediction device 20. The in-vehicle device 13 is connected to the traffic situation prediction device 20 via the radio base station 11 and the network 10. The traffic situation prediction apparatus 20 is also connected via a network 10 to a computer 14 of a traffic information center.

The traffic information center is an external organization that provides traffic information such as a VICS center. Here, the computer 14 that is managed and operated generates traffic information at the present time, and sends it to the traffic situation prediction device 20 and the in-vehicle device 13. Deliver to
The traffic information is distributed in units of areas managed by each management organization. For example, traffic information related to general roads is divided into prefectures, and traffic information related to expressways is divided into administrative offices such as the branches of the Japan Highway Public Corporation (JH), and public corporations such as the Metropolitan Expressway Public Corporation and the Hanshin Expressway Public Corporation. Divided and delivered. The management organization, which is a unit for distributing this traffic information, is hereinafter referred to as an information source.

The traffic information distributed by the traffic information center includes traffic jam information, travel time information, event regulation information, obstacle information, parking lot information, SA / PA information (SA: service area, PA: parking area), text message information, and the like. Yes, it is distributed for each information type. These traffic information includes provision time. Since the distributed traffic information represents the traffic situation at the current time, a time slightly earlier than the time at which the traffic information distributed as the provision time was received is received.
That is, the traffic information distributed here is current traffic information and does not include future traffic information (predicted traffic information). Each of the traffic information described above includes a code indicating an information source (information source code). In addition, the traffic information is basically a link unit, and in fact, the mesh area identification code (mesh code) obtained by dividing the area on the map, the link type identification code (link classification), It is distributed together with the time (current time) created for each combination of three codes of link identification codes (link numbers). Hereinafter, the combination of the three codes is simply referred to as a link.

The in-vehicle device 13 requests the traffic condition prediction device 20 to download the traffic information related to the specified time and mesh (or link), or the shortest route information related to the specified departure place, destination and departure date and time, The predicted traffic information or the shortest route information transmitted from the traffic condition prediction device 20 is received. The in-vehicle device 13 may measure information (probe information) such as the position and speed of the vehicle 12 and transmit the measurement information to the traffic condition prediction device 20. Moreover, you may provide the function to receive the information transmitted from a traffic information center (computer 14).
Then, the received information is superimposed on the map screen displayed on the display, or converted into a simple figure and displayed. Furthermore, a route search function and a route guidance function between the departure point (current location) and the destination are also provided.

The traffic situation prediction device 20 is configured in hardware from a storage device, a calculation device, an input / output device, and a communication device (all not shown).
The traffic situation prediction device 20 receives current traffic information from the traffic information center (computer 14) via a communication device, and the computing device generates future traffic information (predicted traffic information) using a storage device. Further, in response to a request from the in-vehicle device 13, for example, a request for predicted traffic information regarding a specified mesh at a specified predicted destination time, or a request for shortest route information at a specified departure date, departure place, and destination, predicted traffic It also has a function of acquiring information and shortest path information and distributing the information to the requested in-vehicle device 13 via a communication device.

  The predicted traffic information is handled in units of links, like the traffic information handled by the traffic information center. FIG. 2 shows an example of a predicted traffic information distribution format. In FIG. 2, “information creation time” represents the time when the predicted traffic information is created, and “prediction destination time number” represents the maximum prediction destination time, and at the same time means the number of repetitions of the prediction destination time information. “Prediction destination time information” represents information of each prediction destination time, and is “prediction destination time”, “number of mesh information”, and “mesh information”. “Predicted time” represents the current or future time. The “mesh information”, “mesh code”, “number of link information”, and “link information (link category, link number, congestion degree, congestion length, travel time)” are the “predicted destination time” described above. Shows predicted traffic information. “Number of mesh information” represents the number of repetitions of “mesh information” in the data sequence. The “link classification” included in the link information indicates a distinction between an expressway and a general road. When the current traffic information is distributed according to the format shown in FIG. 2, the prediction destination time may be the same as the information creation time or the current time.

  Returning to the schematic configuration diagram of the traffic information system in FIG. A traffic situation prediction apparatus 20 according to an embodiment of the present invention includes a data / request reception unit 21, an information reception / accumulation unit 22, a traffic situation prediction unit 23, a prediction information distribution unit 24, a route search unit 25, and an accumulation. It consists of DB30, statistics DB31, prediction DB32, and map DB33.

The data / request receiving unit 21 receives a request and probe information from a user of the in-vehicle device 13 and receives current status information transmitted from the traffic information center (computer 14). Then, according to the received contents, the information is distributed and transferred to the information receiving / accumulating unit 22 and the predicted information distributing unit 24.
The request from the user of the in-vehicle device 13 includes a request for downloading predicted traffic information for each mesh at a specified predicted destination time, a request for shortest route information, and the like. The request from the user is transferred to the prediction information distribution unit 24, and the probe information from the in-vehicle device 13 and the current traffic information from the traffic information center (computer 14) are transferred to the information receiving / accumulating unit 22. The information receiving / accumulating unit 22 accumulates the probe information transferred via the data / request receiving unit 21 and the current traffic information in the accumulation DB 30.

The storage DB 30 is a database in which the probe information transmitted from the in-vehicle device 13 and the current traffic information transmitted from the traffic information center (computer 14) are sequentially stored. Used.
The statistics DB 31 corresponds to the day type (day of the week, weekday / Saturday / holiday, fifty days, weather, etc.) based on the traffic information for each link accumulated in the past in the traffic situation prediction unit 23. This is a database in which the data created here is stored. For example, when link travel time is targeted as traffic information, it is assumed that there are multiple past days corresponding to the day type of the prediction target day, and the average value, median value, etc. of link travel time for those same time zones The statistical value corresponds to the statistical travel time of the link. Similarly, in the case where link traffic congestion information such as traffic jam length and traffic congestion level is targeted as traffic information, statistical values such as average values for each data related to the same time zone correspond to the statistical traffic congestion information of the link. That is, the statistics DB 31 is used to search travel time and traffic jam information using links, day types, and times as keys.

When the information receiving / accumulating unit 22 receives the current traffic information transmitted from the traffic information center (computer 14), the traffic status prediction process by the traffic status prediction unit 23 is executed subsequently. The traffic situation prediction unit 23 reads the present (present situation) and past traffic information accumulated in the accumulation DB 30, and further reads data stored in the statistics DB 31 that is statistical traffic information, in accordance with a traffic situation prediction method described later. Predict the traffic situation for the predicted destination time (future). Then, based on the predicted traffic situation, the predicted traffic information shown in FIG. 2 is generated and output to the prediction DB 32. Details of the processing in the traffic situation prediction unit 23 will be described later.
The prediction DB 32 is a database in which predicted traffic information generated by the traffic situation prediction unit 23 is stored. An example of data included in the prediction DB 32 is as shown in FIG. The prediction DB 32 is accessed by the prediction information distribution unit 24 and the route search unit 25, where necessary predicted traffic information is searched and output.

The prediction information distribution unit 24 receives a request (prediction / route data request) from the user of the in-vehicle device 13 via the data / request reception unit 21. When the received user request is a download of predicted traffic information, the prediction information distribution unit 24 searches the prediction DB 32 using the specified time and the specified mesh (or specified link) included in the user request as input, and is extracted here. The predicted traffic information is converted into predetermined download format data and transmitted to the in-vehicle device 13.
When the received user request is a download of the shortest route information, the prediction information distribution unit 24 transmits the data of the departure place, the destination, and the departure date and time included in the user request to the route search unit 25, and the route search unit 25 The shortest path information related to the transmission data is received from the data, and the received shortest path information is converted into predetermined download format data and transmitted to the in-vehicle device 13.

The route search unit 25 refers to the predicted traffic information in the prediction DB 32 and the link information included in the map DB 33 based on the departure point, the destination, and the departure date and time transmitted from the prediction information distribution unit 24, and the shortest route. A search process is performed, and the obtained shortest path information is transmitted to the prediction information distribution unit 24. The shortest route information described above includes link information from the departure point to the destination and total travel time (or estimated arrival time).
The map DB 33 stores, for each mesh code, the link data of each link constituting the road included in the mesh area. The link data indicates coordinate information of two nodes (start node and end node) constituting a link for each link number, road type information including the link, link length information indicating the link length, and link speed limit. It includes regulated speed information, link numbers of links connected to the two nodes, and the like. Here, by distinguishing the start node and the end node for the two nodes constituting the link, the upward direction and the downward direction of the same road are managed as different links. The map data also includes information on road structures other than the roads included in the corresponding mesh area (for example, name, type, coordinate information, etc.).

FIG. 3 is a block diagram showing a functional development of the internal configuration of the traffic situation prediction unit in the traffic situation prediction apparatus according to the embodiment of the present invention.
As shown in FIG. 3, the traffic situation prediction unit 23 includes a traffic information acquisition unit 231, a prediction calculation selection processing unit 232, a highway traffic situation prediction calculation processing part 233, and a general road traffic situation prediction calculation process. Unit 234 and a traffic jam information generation output unit 235. Note that the storage DB 30, the statistics DB 31, the prediction DB 32, and the map DB 33 shown in FIG. 3 are the same as the respective databases shown in FIG.

The traffic information acquisition unit 231 takes in current and past traffic information from the traffic information center (computer 14) and the accumulation DB 30 or the statistics DB 31, respectively, a prediction calculation selection processing unit 232, and a traffic situation prediction calculation for highways. It has the function to transfer to the process part 233 and the traffic condition prediction calculation process part 234 for general roads.
The prediction calculation selection processing unit 232 refers to the information about the information source included in the current traffic information and the link classification, and when the link classification corresponds to one of an expressway, a toll road, and an automobile exclusive road, the expressway The traffic condition prediction calculation processing unit 233 performs a prediction calculation process. If none of the conditions applies, the road traffic condition prediction calculation processing unit 234 has a function of executing a prediction calculation process.

The traffic situation prediction calculation processing unit 233 for highways is associated with one or more propagation source links associated with each prediction target link, and for each propagation source link, the traffic situation propagation phenomenon composed of a propagation time and a weighting coefficient is detected. A prediction parameter constituting the used prediction model is generated (second step), a future traffic situation is predicted and calculated based on the prediction model using the traffic information and the prediction parameter, and the predicted traffic situation is output. (Third step) A function is provided.
For this reason, the traffic condition prediction calculation processing unit for highways 233 includes a prediction link information file 330, a prediction link information generation unit 331, a similarity calculation unit 332, a prediction parameter generation unit 333, and a prediction parameter information file 334. , A prediction parameter acquisition unit 335, a traffic jam propagation type prediction determination unit 336, a predicted traffic situation generation output unit 337, and a predicted traffic situation correction processing unit 338.

The predicted link information generation unit 331 searches the map data including the link attribute information for a link (n ≧ 1) that is n-order connected to the predicted link, and among the links searched here, the predicted link information is determined from the predicted link. A link corresponding to at least one of a link within a distance, a link whose link length is longer than a predetermined length, a link whose main link type is a main line, and a link for which traffic information is provided is extracted as a propagation source link, and a predicted link It has a function of storing in the information file 330 and transferring it to the similarity calculation unit 332.
The similarity calculation unit 332 calculates the similarity between the traffic information of the prediction target link and the traffic information obtained by shifting the traffic information of the propagation source link in the past by the propagation time prepared for each, thereby generating a prediction parameter. The function of transferring to the unit 333 is provided.

The prediction parameter generation unit 333 selects the propagation time with the highest similarity calculated by the similarity calculation unit 332 as the propagation time of the propagation source link, and based on the traffic information of the prediction target link and the selected propagation time. A function of calculating the weighting coefficient and intercept that minimize the error of each traffic information, generating a prediction parameter, and storing it in the prediction parameter information file 334. Have.
The prediction parameter acquisition unit 335 has a function of acquiring the prediction parameter generated by the prediction parameter generation unit 333 from the prediction parameter information file 334 and transferring the prediction parameter to the traffic jam propagation type prediction determination unit 336 when the traffic situation is predicted.

The traffic jam propagation type prediction determination unit 336 determines whether or not the traffic jam propagation type prediction described later can be executed from the prediction parameter fetched via the prediction parameter acquisition unit 335 and the past traffic information, and the prediction fetched first. It has a function of transferring the parameters to the predicted traffic situation generation output unit 337 or to a traffic situation prediction calculation processing unit 234 for general roads described later.
When the predicted traffic situation generation output unit 337 determines that the ratio of the links determined that the travel speed is higher than the predetermined free traveling speed among the one or more propagation source links exceeds the predetermined value, The current traffic information is output to the predicted traffic situation correction processing unit 338 as predicted traffic information. In addition, when it is determined that the ratio of links determined that the travel speed is higher than the predetermined free running speed is equal to or less than the predetermined value, the current information and the prediction target time are used using the traffic information and the prediction parameters. A function of obtaining an expected travel speed at an expected destination time indicating a time width of the target travel time, calculating a predicted travel time by dividing the link length included in the predicted link by the predicted travel speed, and transferring the predicted travel time to the predicted traffic condition correction processing unit 338. Have.

  The predicted traffic condition correction processing unit 338 determines the weighting factor included in the prediction parameter, the statistical information obtained by statistically processing the past traffic information, and the weighting factor in the current traffic information, which are determined according to the prediction destination time. Each of the travel times calculated by each of them is subjected to a weighted average process, and the predicted travel time for each prediction destination time is corrected and stored in the prediction DB 32.

On the other hand, the traffic condition prediction calculation processing unit 234 for general roads takes in statistical traffic information obtained by statistically processing past traffic information, and quantitatively indicates the similarity between current and past traffic information and statistical traffic information. It has a function of calculating a divergence degree, predicting and calculating a future traffic situation using the calculated divergence degree and statistical traffic information, and outputting the predicted situation.
For this reason, the traffic condition prediction calculation processing unit 234 for general roads includes a statistical traffic information acquisition unit 341, a statistical prediction determination unit 342, a divergence degree calculation unit 343, and a predicted traffic condition generation output unit 344. .

The statistical traffic information acquisition unit 341 has a function of capturing statistical traffic information obtained by statistical processing of past traffic information and transferring the statistical traffic information to the statistical prediction determination unit 342.
The statistical-type prediction determination unit 342 determines that the past statistical information is valid for a predetermined ratio or more and the statistical traffic information of the prediction destination time is valid, and activates the calculation of the deviation degree by the deviation degree calculation unit 343. It has a function to do. The divergence degree calculation unit 343 has a function of calculating a divergence degree that quantitatively indicates the similarity between the current and past traffic information and the statistical traffic information, and the data regarding the divergence degree calculated here is a predicted traffic situation generation. The data is transferred to the output unit 344.

The predicted traffic situation generation output unit 344 predicts and calculates the future traffic situation using the deviation degree output by the deviation degree calculation unit 343 and the statistical traffic information output via the statistical traffic information acquisition unit 342. It has a function to output the predicted traffic information generated as a result.
The predicted traffic situation generation output unit 344 compares the degree of divergence with the threshold and corrects the degree of divergence according to the predicted destination time indicating the time span from the current time to the prediction target time and the statistical traffic information. When the future traffic situation is predicted and calculated by adding together with the difference, the weight of the statistical traffic information and the current traffic information determined according to the predicted destination time when the degree of divergence is judged to be large by comparing with the threshold The coefficient is added to each of the statistical traffic information and the current traffic information to calculate the future traffic situation, and each is stored in the prediction DB 32.

  The traffic jam information generation / output unit 235 displays the current traffic information and the traffic status output as a result of the prediction calculation by the traffic condition prediction calculation processing unit 233 for highways or the traffic condition prediction calculation processing unit 234 for general roads. Used, quantifying the degree of traffic jam at each forecast destination time indicating the time span from the current time to the prediction target time as a change rate compared with the current time, and comparing the change rate with a threshold value and pre-defined traffic jam trend information It has a function to select and output.

FIG. 4 is a hardware block diagram showing the internal configuration of the in-vehicle device in the traffic information system shown in FIG.
As shown in FIG. 4, the in-vehicle device 13 includes an arithmetic processing unit 40, a display 41, a data storage device 42, a voice input / output device 43, an input device 44, a wheel speed sensor 45, and a geomagnetic sensor 46. A gyro sensor 47, a GPS (Global Positioning System) receiver 48, an in-vehicle LAN (Local Area Network) device 49, a network communication device 50, an FM multiplex broadcast receiver 51, and a beacon communication device 52. Is done.

The arithmetic processing unit 40 is a CPU (Central Processing Unit) serving as a control center of the in-vehicle device 13. The arithmetic processing unit 40 obtains the current position based on, for example, information output from the wheel speed sensor 45, the geomagnetic sensor 46, the gyro sensor 47, and the GPS receiver 48. Further, map data necessary for display is read from the data storage device 42 based on the obtained current position information. Further, the read map data is developed into graphics, and a mark (icon) indicating the current position is superimposed thereon and displayed on the display 41.
Also, using the map data and statistical traffic information stored in the data storage device 42, the predicted traffic information received from the traffic status prediction device 20, or the current traffic information received from the traffic information center (computer 14), An optimum route (recommended route) connecting the current position (departure point) set by the user and the destination is searched. The recommended route is guided to the user using the voice input / output device 43 and the display 41.

  In addition, you may make it perform each function which the traffic condition prediction apparatus 20 mentioned later has with the arithmetic processing part 40 of the vehicle-mounted apparatus 13, and the vehicle-mounted apparatus 13 does not connect to the traffic condition prediction apparatus 20 in that case. Alternatively, it is necessary to connect to the computer 14 of the traffic information center. As connection means at this time, business radio such as taxis, mobile radio communications such as mobile phones and wireless LANs, or broadcast radio communications such as satellite communications, FM multiplex broadcasting, and terrestrial digital broadcasting can be considered.

The display 41 is a man-machine interface that displays graphics information generated by the arithmetic processing unit 40, and includes a CRT (Cathode Ray Tube), a liquid crystal display, or the like. Signals between the arithmetic processing unit 40 and the display 41 are generally connected by RGB signals or NTSC (National Television Standards Committee) signals.
The data storage device 42 includes a recording medium such as a CD-ROM, DVD-ROM, HDD, or IC card, and a reading or writing device for these recording media. These recording media store various data such as map data and statistical traffic information.

By the way, map data contains the link data of each link which comprises the road contained in the mesh area | region for every mesh code | cord | chord. The link data includes coordinate information of two nodes (start node and end node) constituting the link for each link number, road type information including the link, link length information indicating the link length, and link speed limit. The restricted speed information indicating the link number of the link connected to each of the two nodes.
Here, for the two nodes constituting the link, the start node and the end node are distinguished from each other so that the up and down directions of the same road are managed as different links. The map data also includes information on road structures other than the roads included in the corresponding mesh area (for example, name, type, coordinate information, etc.).

The voice input / output device 43 converts a message to the user generated by the arithmetic processing unit 40 into a voice signal and outputs the voice signal. In addition, it recognizes the voice uttered by the user and performs processing for transferring the contents to the arithmetic processing unit 40.
The input device 44 is a man-machine interface that captures an instruction from the user, and includes a hardware switch such as a scroll key and a scale change key, a joystick, a touch panel mounted on the display 41, a voice input microphone, and the like. The

The wheel speed sensor 45, the geomagnetic sensor 46, the gyro sensor 47, and the GPS receiver 48 are used by the in-vehicle device 13 to detect the current position (own vehicle position).
The wheel speed sensor 45 measures the travel distance from the product of the wheel circumference and the measured number of rotations of the wheel, and further measures the angle bent from the difference in the number of rotations of the paired wheels. Further, the geomagnetic sensor 46 detects the magnetic field held by the earth and detects the direction in which the moving body is facing. Further, the gyro sensor 47 is constituted by an optical fiber phylo, a vibration gyro, or the like, and detects an angle at which the moving body rotates.
The GPS receiver 48 receives a signal from a GPS satellite, and measures the distance between the mobile body (vehicle 12) and the GPS satellite and the rate of change of the distance between three or more satellites. Measure the current position, travel speed, and travel direction.

The in-vehicle LAN device 49 receives and monitors various information of the vehicle 12 on which the in-vehicle device 13 is mounted, for example, door opening / closing information, light lighting state information, engine status and failure diagnosis result. The network communication device 50 is a device that mediates transmission / reception of information between the in-vehicle device 13 and the traffic situation prediction device 20. The network communication device 50 includes a wide-area communication device such as a mobile phone, a PHS (Personal Handyphone System), and a wireless LAN.
The in-vehicle device 13 accesses the traffic condition prediction device 20 periodically or in response to a request from the user via the network communication device 50, receives the predicted traffic information and the shortest route information, and stores it in the data storage device 42. Remember.

The FM multiplex broadcast receiver 51 transmits the current traffic information from the traffic information center (computer 14) to the FM broadcast station, and outline traffic information (such as graphics) and characters transmitted by the FM broadcast station as an FM multiplex broadcast signal. Receive information.
Further, the beacon communication device 52 transmits the current traffic information from the traffic information center (computer 14) to the traffic control center in each region, and the detailed current status that the traffic control center transmits via a beacon that is a communication device on the road. Receive traffic information (link travel time information, etc.).

5 to 18 are diagrams cited for explaining the operation of the traffic information system including the traffic condition prediction apparatus according to the embodiment of the present invention.
Hereinafter, the operation of the traffic information system including the traffic condition prediction apparatus according to the embodiment of the present invention will be described in detail with reference to FIGS.

FIG. 5 is a flowchart cited for explaining the route search processing procedure by the route search unit 25 of the traffic condition prediction apparatus 20.
In the flowchart shown in FIG. 5, the traffic situation prediction device 20 first starts from the in-vehicle device 13 via the wireless base station 1, the network 10, and the internal data / request reception unit 21 and the prediction information distribution unit 24. The transferred data relating to the departure point, destination, and departure date and time is set in the route search unit 25 (step S50). Thereby, the route search unit 25 specifies each mesh code included in the route search region including the departure point and the destination from the coordinates of the departure point and the destination.

Here, an example of specifying each mesh code included in the route search area will be described. Here, a rectangular mesh having both ends of the mesh including the starting point and the destination is taken as a rectangular mesh that is one size larger. For example, when a rectangular mesh having both ends of a mesh including the starting point and the destination is an m × n mesh, the search area is (m + 2) × (n + 2) mesh.
Next, the route search unit 25 reads link data registered in the map DB 33 corresponding to the identified mesh code (step S51). Subsequently, the route search unit 25 obtains predicted traffic information (predicted travel time) of each candidate link after the departure date and time for each mesh code by searching the prediction DB 32 based on the identified mesh code and departure date and time. (Step S52).

Further, the route search unit 25 uses the link data read in advance, and uses the link whose start node is the end node of the link extracted from the heap table (hereinafter referred to as an extracted link) in step S57 described later as the shortest route. Are selected as candidate links (hereinafter referred to as candidate links) (step S53).
However, when the process in step S57 is not performed, that is, in the initial stage where the link is not registered in the heap table, instead of selecting a link having the end node of the extracted link as a start node as a candidate link, departure At least one link where the ground exists or is close to the departure place is selected as a candidate link (step S53).

Here, the heap table is a memory area used by an OS (Operating System) or application software. Here, the link data of the candidate link is specifically calculated from the starting point to the end node of the candidate link. It is a table for registering together with (total travel time).
FIG. 6 shows an example of the data structure of the heap table in a table format. As shown in FIG. 6, a record (corresponding to one row of the table) is registered for each candidate link in the heap table. The record includes the link number of the candidate link, the cost (travel time) of the candidate link, the link number of the connection source link where the end node connects to the start node of the candidate link, the total cost from the departure point to the candidate link, and In step S57, each field for registering each data of an extraction flag indicating whether or not it has been set as an extraction link.

Next, the route search unit 25 calculates an expected arrival time at the end node of the extracted link (step S54). This can be calculated by identifying the record of the candidate link currently set for the extraction link from the heap table and adding the total cost registered in the field of this record at the departure time. Then, a record including the candidate link is specified for each candidate link from the traffic information by link. Then, the link travel time in the time zone (attention time zone) to which the estimated arrival time belongs is specified from the specified record.
Next, the route search unit 25 calculates the total cost from the departure place to the candidate link for each candidate link. Specifically, the travel time in the attention time zone of the candidate link is added to the total travel time of the extracted links registered in the heap table, and the addition result is used as the total cost of the candidate links. Then, a record for each candidate link is added to the heap table. Subsequently, the link number of the corresponding candidate link, the cost of the candidate link, the link number of the extracted link (connection source link), and the total cost from the departure place to the candidate link are registered in each field of each added record. In the extraction flag field, an extraction flag “not yet” indicating that the extraction link has not yet been set is registered (step S55).

Next, the route search unit 25 includes a destination link or a link close to the destination (hereinafter referred to as a destination link) among the candidate links newly added to the heap table in the process of step S55 performed immediately before. ) Exists (step S56).
In the process of step S56, if the destination link does not exist, the route search unit 25 next sorts the records of the candidate links registered in the heap table in ascending order of the total cost, and the extraction flag field is Among the candidate links that are “unextracted” (hereinafter referred to as unextracted links), the unextracted link with the lowest total cost is extracted from the heap table, for example, by extracting the unextracted link located at the top. Then, the current extracted link is changed to the unextracted link, and the extraction flag of the unextracted link is changed from “not yet” to “done” (step S57). Then, the process returns to S53.

On the other hand, in the process of step S56, the route search unit 25 performs the shortest route determination process when there is a destination link. Specifically, the connection link of the destination link (the link registered in the connection source link number field in the record of the destination link) is searched from the heap table, and the link (the shortest path is configured) of the searched link ( Hereinafter, it is referred to as a configuration link. Next, the route search unit 25 searches whether there is a connection source link of the configuration link. If there is a connection source link, the route search unit 25 determines this connection source link as the configuration link, and there is the connection source link. Further search whether or not.
The route search unit 25 repeats the above-described processing until there is no connection source link of the configuration link, that is, until the configuration link becomes a departure place or a link close to the departure place (hereinafter referred to as a departure place link), Each constituent link constituting the shortest path is determined. Then, as the shortest route information, the link data of each constituent link constituting the shortest route and the predicted traffic information (predicted travel time) obtained by the processing in step S52 are transferred to the predicted information distribution unit 24 together with the departure time information. (Step S58).

As described above, the route search unit 25 performs the route search process using the predicted traffic information, so that the route with the shortest time avoiding the traffic jam predicted in advance can be distributed.
The in-vehicle device 13 also has the same functions and data as the route search unit 25, the prediction DB 32, and the map DB 33 described above, or by downloading the predicted traffic information from the traffic situation prediction device 20, or It is possible to obtain the predicted traffic information of the future by reading the statistics DB 31 provided in the device 13, and to provide the shortest time route that avoids the traffic jam predicted in advance by executing the route search process described above. it can.

(Overall traffic condition prediction process)
FIG. 7 is a flowchart showing an overall flow of traffic situation prediction processing by the traffic situation prediction unit 23 of the traffic situation prediction apparatus 20.
In the flowchart shown in FIG. 7, the information receiving / accumulating unit 22 first reads the current traffic information transmitted from the traffic information center (computer 14) and transfers it to the traffic status predicting unit 23 (step S70). Subsequently, the traffic condition prediction unit 23 determines whether the target road of the traffic information is a highway or a general road from the information source code included in the transferred current traffic information (step S71). The determination is made by referring to the data of the link classification that determines whether the road is a highway or a general road for each information source. Here, the expressway refers to a case corresponding to any of a highway, a toll road, and an automobile exclusive road, and a general road refers to a road not corresponding to any of them.

When the traffic situation prediction unit 23 determines that the road is an expressway in the process of step S71, the traffic situation prediction unit 23 executes a prediction process for the expressway (step S72). Details of the highway prediction process will be described later. Further, when the traffic situation prediction unit 23 determines that the road is a general road in the process of step S71, the traffic situation prediction unit 23 performs a prediction process for the general road (step S73). Details of the general road prediction process will be described later. Finally, the predicted traffic information is stored in the prediction DB 32 and is output to the in-vehicle device 13 via the network 10 via the prediction information distribution unit 24. In addition, the traffic information to be distributed includes traffic jam trend information described later as required (step S74). Thus, the traffic situation prediction process ends.
Here, the reason for making the prediction processing method different for the expressway and the general road is to improve the prediction accuracy in consideration of the characteristics. For example, statistical traffic information tends to have relatively high accuracy on ordinary roads, but accuracy on highways may deteriorate due to being greatly affected by sudden events such as accidents compared to ordinary roads. Many. On the other hand, on a highway, branch points for avoiding traffic jams are limited to ramps (interchanges) and junctions. Therefore, when traffic jams occur, the roads tend to extend greatly. The main object of the present invention is to realize highly accurate traffic situation prediction in consideration of this characteristic.

The details of the highway prediction process (step S72) and the general road prediction process (step S73) will be described below with reference to the flowcharts shown in FIGS.
First, prior to the detailed description of the highway prediction process, the internal configuration functional block of the traffic condition prediction unit 23 (highway traffic condition prediction calculation processing unit 233) shown in FIG. This will be described in detail with reference to the flowcharts shown in FIGS.

(Generation of prediction parameters for expressways)
Although the prediction parameter for highways should be updated regularly using the past traffic information accumulate | stored in accumulation | storage DB30, the update period changes with parameters. For example, the predicted link information such as the link number, link length, restriction speed, and propagation source link that can be generated only from the link information stored in the map DB 33 may be updated according to the update time of the map DB 33. About once or twice a year.
On the other hand, prediction parameter information such as various parameters (intercept, coefficient, propagation time) according to free travel speed and prediction destination time should be generated according to past traffic information, so a certain update frequency is required. For example, the update frequency may be about once a week or once a month. Therefore, after the prediction link information is generated, the prediction parameter information is generated.

First, prediction link information generation processing will be described with reference to the flowchart shown in FIG. First, the predicted link information generation unit 331 reads link information included in the map DB 33, and extracts a road link corresponding to any one of an expressway, a toll road, and an automobile road as a prediction target link (step S80). The prediction target links may be further narrowed down to links for which traffic information is provided by reading the storage DB 30.
Next, the predicted link information generation unit 331 uses the link connection information included in the link information, and the link that is primarily connected to the prediction target link and the link that is further connected (prediction target link). The n-th order connection link (n ≧ 2)) and the like are sequentially searched as propagation source link candidates (step S81).

Subsequently, the predicted link information generation unit 331 determines whether or not the condition of the propagation source link is established for each propagation source link candidate searched in step S81 (step S82). If the condition is satisfied (YES), the predicted link information generation unit 331 proceeds to the process of step S83, and if not satisfied, returns to step S81 to search for another propagation source link.
Regarding the conditions for the determination, for example, the link length is not short (for example, 100 m or more), the road is a main road (not a side road, a ramp, or a crossover), the distance from the prediction target link is not more than a predetermined distance, Satisfying at least one of the above. The propagation source links may be further narrowed down to links for which traffic information is provided by reading the storage DB 30.

If it determines with YES by the process of step S82, the prediction link information generation part 331 will set the propagation source link candidate to a propagation source link, and will transfer to the similarity calculation part 332 (step S83).
The predicted link information generation unit 331 determines whether or not the above-described processing has been completed for all the prediction target links extracted in step S80 (step S84), and if completed (YES), the process of step S85 is performed. Proceed to the process, and if not completed, repeat the processes of steps S81 to S83 described above until the process ends. If the predicted link information generation unit 331 determines YES in step S84, the predicted link information file outputs the correspondence relationship between the prediction target link and the propagation source link, the link length, the restriction speed, and the like to the similarity calculation unit 332 as a predicted link information file. The process is terminated (step S85).

Next, prediction parameter information generation processing will be described with reference to the flowchart shown in FIG.
The similarity calculation unit 332 first reads various types of information necessary for generating prediction parameter information (step S90). The information read here is a predicted link information file generated by the predicted link information generation unit 331 and past traffic information stored in the accumulation DB 30. The number of days required for past traffic information may be, for example, the latest one week, but it should be noted that the latest propagation tendency can be more accurately grasped by targeting the latest data as much as possible. .

Then, the similarity calculation unit 332 calculates the similarity between the traffic information (travel speed) of the prediction target link and the traffic information (travel speed) of the propagation source link shifted in the past by the propagation time (step S91). . The similarity is calculated by calculating a correlation coefficient between two data of the traffic information of the prediction target link and the traffic information of the propagation source link, and the square root of the average value of the square error between the two data. An RMS (Root Mean Square) error can be considered. Here, the propagation time is given in increments of 1 minute from 0 minutes to 180 minutes, for example, in the process of step S92 described later.
Next, the similarity calculation unit 332 determines whether or not the process of step S91 has been completed for all propagation times (step S92), and if completed (YES), the similarity calculated to the prediction parameter generation unit 333 is determined. The data relating to the degree is transferred and the process proceeds to the process of step S93. If not completed (NO), the propagation time is changed until the process is completed, and the process of step S91 is repeated.

When the prediction parameter generation unit 333 determines YES in the process of step S92, the prediction parameter generation unit 333 sorts the propagation times in descending order of similarity by comparing the similarities regarding all the propagation times. Here, when the method of calculating the correlation coefficient is adopted as the similarity calculation method, the similarity is higher as the similarity value is larger (closer to 1), and the method of calculating the RMS error is adopted. It is determined that the similarity is higher as the similarity value is smaller (closer to 0).
Next, the prediction parameter generation unit 333 determines an optimal propagation time for each prediction destination time k based on the sorted propagation times (step S93). For example, assume that the propagation time with the highest similarity is τ1. Here, when the prediction destination time k is equal to or less than τ1, τ1 is determined as the propagation time of the propagation source link at the prediction destination time k. However, if the predicted destination time k is larger than the τ1, the comparison is made with the propagation time τ2 having the second highest degree of similarity, and the search is continued until a propagation time that makes the predicted destination time k lower than that is found. Hereinafter, the traffic information of the propagation source link shifted in the past by the propagation time determined in this way is simply referred to as traffic information of the propagation source link.

The prediction parameter generation unit 333 cooperates with the similarity calculation unit 332 to repeat the processes of steps S91 to S93 for all propagation source links in the prediction target link (S94).
Subsequently, the prediction parameter generation unit 333 performs propagation source link exclusion processing for each prediction destination time (step S95). In this exclusion process, the propagation source link q determined to be absolutely low in similarity is predicted by comparing the degree of similarity of the determined propagation source link with respect to the traffic information and a predetermined threshold value set in advance. It is excluded from the propagation source link used for. That is, the weighting coefficient bq (k) of the propagation source link q at the prediction destination time k that is a prediction parameter is set to zero. Further, when a propagation source link pair exceeding a predetermined threshold is found by calculating a cross-correlation coefficient between propagation source links, one of them may be similarly excluded. This is one means for improving the accuracy of calculation of a prediction parameter (multiple regression analysis) described later.

Next, the prediction parameter generation unit 333 performs weighting of each propagation source link so that the error of the predicted travel time is reduced by performing multiple regression analysis using the travel speed of the propagation source link remaining in the exception processing described above. A prediction parameter including a coefficient and an intercept is calculated (step S96).
The prediction parameter generation unit 333 further repeats the processing from steps S91 to S96 described above for all prediction destination times (S97). Moreover, the process from step S91 to step S97 is repeated for all prediction target links (step S98). Finally, the calculated prediction parameter is output to the prediction parameter information file 334 (step S99).

An example of the data format of the prediction parameter is shown in FIG. As shown in FIG. 12, the prediction parameter is managed for each mesh (information source number) of the information source, and further managed for each prediction target link (prediction target link number). Here, the “number of mesh information” means the number of repetitions of mesh information in the data arrangement, and the prediction destination time number (n) represents the number of repetitions of the prediction destination time in the prediction target link information. Further, “mesh information size” represents the data size (unit: byte) of mesh information, and “number of intra-mesh prediction target links” represents the number of repetitions of prediction target link information. In the prediction target link, link number, a link length, is stored attribute data of the link such as speed limits (speed limit), and further, links the vicinity of propagating traffic to the link (hereinafter, propagation source link ^#. 1 to ^# m Information) is also stored.
The information of the propagation source link includes a mesh code, a link number, a link length, and a free travel speed (corresponding to a travel speed when the traffic is a free stream). In the prediction target link information further prediction target time ^# 1 to various prediction parameters for each ^# n (intercept _{b 0,} the coefficient _b 1 ~b _m, the propagation time τ ₁ ~τ _m) is included. Here, the prediction destination time means a future time to be predicted, for example, 5 minutes, 10 minutes,..., 180 minutes, or the like.

(Prediction processing for expressways)
Next, with respect to the highway prediction process using the prediction parameter generated as described above, the internal configuration functional block diagram of the traffic situation prediction unit 23 (highway traffic situation prediction calculation processing unit 233) shown in FIG. This will be described in detail with reference to the flowchart shown in FIG.
In the flowchart shown in FIG. 10, the traffic situation prediction calculation processing unit 233 for highways reads the above-described prediction parameters, and the traffic information acquisition unit 231 traffics from the accumulation DB 30 in a period retroactive to the past for a predetermined period from the present. Information is read and transferred to the highway traffic situation prediction calculation processing unit 233 (step S100). Here, the prediction parameter is created in advance based on past traffic information, and is expanded as a prediction parameter information file 334 on the memory during the traffic situation prediction process. The prediction parameter acquisition unit 335 includes: By referring to the memory (prediction parameter information file 334), the prediction parameter is acquired and transferred to the traffic jam propagation type prediction determination unit 336.

Whether or not the traffic jam propagation type prediction determination unit 336 can execute the traffic jam propagation type prediction from the prediction parameter acquired through the prediction parameter acquisition unit 335 and the past traffic information acquired through the traffic information acquisition unit 231. Is determined (step S101).
Specifically, the traffic jam propagation type prediction determination unit 336 refers to the traffic information of the propagation source link at the past time (t−τ) from the current time t by the propagation time τ based on the propagation time τ of each propagation source link. Here, when it is confirmed that the traffic information of all the propagation source links exists, it is determined that the prediction is possible (YES), and the process proceeds to the traffic jam propagation type prediction process (step S102) by the predicted traffic state generation output unit 337. In cases other than the above, it is determined that the prediction is impossible (NO), and the process proceeds to a general road prediction process (step S73) described later.

If it is determined YES in the process of step S101, the predicted traffic condition generation / output unit 337 first determines whether or not the traffic condition is in a state of maintaining a smooth state (whether the traffic condition changes) (step S102). ).
Here, based on the propagation time τ of each propagation source link, the predicted traffic situation generation output unit 337 refers to the traffic information of the propagation source link at the past time point (t−τ) from the current time t by the propagation time τ, When the link whose travel speed is higher than the free travel speed among all the propagation source links exceeds a predetermined ratio, it is determined that the traffic situation of the prediction target link does not change (NO), and the process of step S104 Execute. On the other hand, in cases other than the above, it is determined (YES) that the traffic situation does not necessarily change (that is, changes), and the process of step S103 is executed.

If YES is determined in step S102, a traffic jam propagation type prediction process is executed, and the predicted traffic situation generation / output unit 337 predicts a travel time after the current time (step S103).
In the prediction processing, the predicted travel speed V (t + k) that is k hours in the future from the current time t is calculated by the following calculation formula (1) using the previously read prediction parameters and past traffic information.

Here, b ₀ is an intercept, and b ₁ (k) is the same as a weighting coefficient _ap described later. Further, v _i is the link travel speed is calculated from the travel time and the link length of the propagation source link i. Then, the predicted travel time is calculated by dividing the link length of the prediction target link by the predicted travel speed V (t + k).
On the other hand, when it is determined NO in the process of step S102, the predicted traffic situation generation / output unit 337 predicts that the traffic situation will remain in the current state (does not change) in the future, regardless of the prediction destination time. The current travel time is set as the predicted travel time without executing the traffic jam propagation type prediction process (step S104).

Next, the predicted traffic situation correction processing unit 338 corrects the predicted travel time calculated from the predicted traffic situation generation / output unit 337 in step S103 or step S104 according to the predicted destination time k (step S105).
In the above correction processing, when the weighting factor of the predicted travel time is a _p , the weighting factor of the current travel time is a _r , and the weighting factor of the statistical travel time is a _s , as shown in FIG. Accordingly, the total of the respective weights is determined to be 1.0, and is developed on a memory table (not shown) of the predicted traffic situation correction processing unit 338. The predicted traffic situation correction processing unit 338 refers to this memory table and corrects the predicted travel time for each predicted destination time by performing a weighted average process on each travel time.

Here, kr in FIG. 13 is a predicted destination time that is a boundary of whether or not the current travel time is used in the correction processing described above, and the current travel time is predicted to be less accurate than the predicted travel time and the statistical travel time. What is necessary is just to set ahead time.
Further, the weighting factor for each travel time may be determined mechanically according to the accuracy characteristics for each link or time zone, but this makes the system complicated, and the traffic information center (computer 14) It can be considered that the weighting factor is temporarily unique due to the influence of noise components included in the traffic information sent and sudden events such as accidents. In order to prevent such a situation, the weight coefficient may be determined empirically or comprehensively. Further, this correction step itself may be omitted.

Returning to the flowchart shown in FIG. 10, the highway prediction process will be described. The predicted traffic situation correction processing unit 338 calculates the predicted travel speed by dividing the link length of the predicted link by the predicted travel time calculated and corrected as described above, and stores it in the prediction DB 32.
On the other hand, the traffic information generation / output unit 235 determines the traffic level of the predicted travel speed by comparing the predicted travel speed read from the prediction DB 32 with a predetermined traffic level determination threshold (step S106). Here, for example, the degree of traffic jam is expressed in three levels: no traffic jam (smooth), congestion (light traffic jam), and traffic jam (heavy traffic jam). For example, in the case of a general road, the congestion degree determination threshold value is congested when the predicted travel speed is less than 10 km / h, congested when the estimated travel speed is less than 10 km / h and less than 20 km / h, and Similarly, in the case of an expressway, the traffic is congested if the predicted travel speed is less than 20 km / h, congested if it is 20 km / h or more and less than 40 km / h, and congested if it is 40 km / h or more. You can set it to none.

The highway traffic condition prediction calculation processing unit 233 determines whether or not the above-described prediction processing has been completed for all prediction destination times k (step S107). Step S106 is repeated, and if completed, the process proceeds to Step S108.
The traffic situation prediction calculation processing unit 233 for the highway finally determines whether or not the prediction process has been completed for all the prediction target links (step S108), and if not, the step S101 is performed for the remaining prediction target links. Repeat the process of step S107. If completed, the highway traffic situation prediction calculation process is terminated.

  As described above, since the traffic situation prediction apparatus 20 can make predictions with relatively simple calculations, it is easy to target a wide area in the whole country without requiring a special computer such as a supercomputer. In addition, since an optimal prediction method is adopted according to the characteristics of the road and the predicted destination time, it becomes possible to predict the future traffic situation with high accuracy. Then, by using the predicted traffic information as the link cost in the shortest route search by the route search unit 25 or the in-vehicle device 13, the route search considering the traffic jam situation in the near future can be performed in advance, and the target can be more reliably and in a short time. A route to reach the ground can be provided.

(Prediction processing for general roads)
Next, regarding the details of the general road prediction process (FIG. 7, step S73), the functional configuration block diagram of the traffic condition prediction unit 23 (traffic road traffic condition prediction calculation processing unit 234) shown in FIG. 3 and FIG. This will be described with reference to the flowchart shown in FIG.

In the flowchart shown in FIG. 11, the traffic information acquisition unit 231 first reads traffic information for a period retroactive from the present to the past for a predetermined period from the accumulation DB 30, and further, statistical traffic information and links corresponding to the prediction target date from the statistics DB 31. The link information composed of the length and the regulated speed is taken in and transferred to the traffic condition prediction calculation processing unit 234 for general roads (step S110). In addition, in the traffic condition prediction calculation processing unit 234 for general roads, an information table (not shown) in which the correspondence relationship between the prediction target date and the day type of the statistical traffic information is predetermined is stored in the memory of the statistical traffic information acquisition unit 341. The statistical traffic information acquisition unit 341 determines the day type of the prediction day with reference to this information table, reads the statistical traffic information corresponding to the determined day type, and sends it to the statistical type prediction determination unit 342. Forward.
The statistical prediction determination unit 342 determines whether or not statistical prediction can be performed from the read past traffic information and statistical traffic information (step S111). The statistical prediction determination unit 342 determines that statistical prediction is possible (YES) if a predetermined ratio or more of the read past traffic information is valid and the statistical traffic information of the prediction destination time is valid. Then, the process proceeds to step S112. In cases other than the above, it is determined that the prediction is not possible (NO), and the process proceeds to step S117 without executing the prediction process.

If YES is determined in the process of step S111, the divergence degree calculation unit 343 calculates the divergence degree between the previously read traffic information (near past) and the statistical traffic information before performing the statistical prediction process. Calculate (step S112).
The degree of divergence D quantitatively represents how similar or far the traffic situation in the near past of the prediction time t is to or far from the traffic situation represented by the statistical traffic information. For example, the following equation ( 2).

  Here, j is a past time up to the past p time. d (t) is the past actual travel speed Vr (t) and the statistical travel speed Vs (t) obtained from the accumulation DB 30, as represented by the following equation (3). Is the difference.

Here, γ is a weighting coefficient (positive real number) of d (t), and is set smaller as j becomes larger. H is a multiplier to be determined in advance, and is a positive real number. That is, the divergence degree represented by the calculation formula (2) corresponds to a weighted average of the difference d (t) between the actual travel speed and the statistical travel speed according to the newness of time.
Further, the divergence degree may be defined by D ′ (t) represented by the following equation (4) instead of the above equation (2).

  However, the calculation formula (4) corresponds to a weighted average of the ratio of d (t) to the actual travel time Tr (t) according to the newness of time.

Returning to the flowchart of FIG. 11, the description of the general road prediction process is continued. The predicted traffic situation generation output unit 344 determines whether the calculated deviation degree is large or small by comparing the deviation degree calculated by the deviation degree calculation unit 343 with a preset deviation degree determination threshold. (Step S113).
When it is determined in step S113 that the degree of divergence is small (YES), the predicted traffic situation generation output unit 344 executes statistical prediction processing with a small degree of divergence (step S114). In the statistical prediction process with a small degree of divergence, the trend is predicted to be similar to the statistical traffic information in the near future because it is close to the traffic situation represented by the statistical traffic information in the near past. For example, the divergence degree D (t ) Is used to calculate a predicted travel speed V (t + k) for the future of k hours from the current time t by the following equation (5).

  However, as shown in FIG. 14, α is a weighting coefficient of the divergence degree D (t) that changes in accordance with the prediction destination time k. Moreover, the prediction calculation formula in the case where D ′ (t) described above is used instead of D (t) as the divergence is, for example, the following calculation formula (6).

  However, as shown in FIG. 15, α ′ is a parameter for correcting the divergence degree D ′ (t) that changes according to the prediction destination time k. In any of the above-described prediction calculation formulas, the predicted travel speed approaches the statistical travel speed Vs as the predicted destination time k becomes further in the future by the parameters α and α ′.

On the other hand, when it is determined in step S113 that the degree of divergence is large (NO), the predicted traffic situation generation output unit 344 executes statistical type prediction processing with a large degree of divergence (step S115).
In the statistical type prediction process with a large degree of divergence, the predicted traffic situation generation output unit 344 is closer to the current traffic information than the statistical traffic information in the immediate future because it is not close to the traffic situation represented by the statistical traffic information in the near past. It is predicted that the trend will be similar to statistical traffic information in the future. Based on this idea, the predicted traffic condition generation / output unit 344 calculates a predicted travel speed V (t + k) of k hours in the future from the current time t by the following arithmetic expression (7).

However, β is a weighting coefficient of the statistical travel speed Vs as exemplified in FIG. Thus, the predicted travel speed approaches the current travel speed Vr as the predicted destination time k is closest, and approaches the statistical travel speed Vs as the future is reached.
Next, similarly to the prediction process for highways, the traffic jam information generation / output unit 235 compares the predicted travel speed calculated in step S114 or step S115 with a predetermined traffic congestion degree determination threshold, thereby predicting the travel Determine the speed congestion. Further, the predicted traffic situation generation / output unit 344 calculates the predicted travel time by dividing the link length of the prediction target link by the predicted travel speed, and stores it in the prediction DB 32 (step S116).

The traffic condition prediction calculation processing unit 234 for general roads determines whether or not the prediction process has been completed for all the prediction destination times k (step S117), and if not, the steps S111 to S11 are performed for the prediction target link. The process of S116 is repeated, and if completed, the process proceeds to step S118.
The traffic condition prediction calculation processing unit 234 for general roads finally determines whether or not the prediction process has been completed for all the prediction target links (step S118), and if not, the step is performed for the remaining prediction target links. The processes from S111 to S117 are repeated. If completed, the general road prediction process is terminated.

(Congestion trend information generation)
Next, the traffic information generation / output unit 235 of the traffic condition prediction unit 23 uses the traffic information predicted by the traffic condition prediction calculation processing unit 233 for highways or the traffic condition prediction calculation processing unit 234 for general roads to predict The operation when the traffic jam tendency information indicating that the level of the traffic jam in the previous time is worsened, relaxed, or not changed compared to the present time is generated, transmitted to the in-vehicle device 13 and displayed is described.
The traffic congestion tendency information generation process is executed by the traffic situation prediction unit 23, and is generated and output according to the following procedure using the traffic situation prediction process result.

First, for the purpose of removing random noise included in the predicted travel time of the target link, the traffic jam information generation / output unit 235 calculates the current travel time at the current time t and the predicted destination time (t + 1) to the predicted destination time (t + n). The smoothing process is performed using each predicted travel time in the near future until. Here, the smoothing process may be executed by a method such as moving average or exponential smoothing, but it is not necessary to execute it when it is not necessary to remove noise.
Subsequently, the current travel time T and the predicted travel time Tp on which the smoothing process has been performed are applied to the following equation (8), and the travel time change rate R (t + k) for the predicted destination time k (1 ≦ k ≦ n): ) Is calculated.

Then, the traffic jam information generation / output unit 235 determines the traffic jam trend information based on the conditional expression of the travel time change rate shown in FIG. Ε1 and ε2 in FIG. 17 are predetermined threshold values, the former being a negative real number and the latter being a positive real number.
The traffic jam tendency information generated here is distributed via the prediction information distribution unit 24, the network 10, and the wireless base station 11 in response to a request from the in-vehicle device 13.

FIG. 18 shows an example of the screen configuration when displaying on the display 41 of the in-vehicle device 13 using the traffic jam tendency information and the current traffic information generated as described above.
In FIG. 18, line segments represent current traffic information (congestion level) divided into three levels (congestion, congestion, and no congestion), and icons represent congestion trend information. Moreover, in FIG. 18, the code | symbol 60 shows the present position of the vehicle equipment 13, and the code | symbol 61 shows the present time. Reference numerals 62 to 70 respectively display a congestion degree and congestion tendency information together. For example, reference numeral 62 represents “traffic jam” that tends to relieve traffic jams, and reference numeral 63 represents “traffic jam” that tends to worsen traffic jams. Similarly, reference numerals 64, 65, 66, and 67 indicate “congestion” that tends to reduce congestion, “congestion” that tends to worsen congestion, “no congestion” that tends to reduce congestion, and worsening congestion. It represents a certain “no traffic jam”. In addition, reference numerals 68 to 70 indicate that the icon of the traffic jam tendency information is not displayed, but this means that the degree of the traffic jam does not change from the current situation. Therefore, reference numerals 68, 69, and 70 indicate “congestion” without change, “congestion” without change, and “no congestion” without change, respectively.

Note that the predicted destination time that is the target of the traffic jam trend information to be displayed on the display 41 is determined by the arithmetic processing unit 40 of the in-vehicle device 13. In determining the predicted destination time, it may be determined by designating a future time to be displayed by the user via the input device 44. As another determination method, a correspondence relationship between the distance from the current position of the in-vehicle device 13 and the predicted destination time is determined in advance, and the predicted destination time of each link may be determined using the correspondence relationship. Good.
In this way, by using the traffic congestion trend information together with the conventional predicted traffic information, it becomes possible to provide more detailed traffic information in consideration of the prediction information (the traffic congestion trend information) from the mere three levels of traffic congestion. For example, if the link ahead of the route is congested at the present time, the driver can know whether the tendency of the congested traffic deteriorates or eases. It is easier to make a decision on detour behavior when the traffic jam trend is worse, and a decision to move forward without changing the route when the traffic jam trend shows mitigation. Will improve.

  As described above, for example, in the case of an expressway, the present invention relates to a step in which a traffic situation prediction device captures current and past traffic information from the outside via a communication device, and a link related to each prediction target link. A step of generating a prediction parameter constituting a prediction model using a propagation phenomenon of a traffic situation composed of a propagation time and a weighting factor for each of the propagation source links, and storing it in a storage device; Predicting a future traffic situation by a calculation device based on the prediction model using the traffic information and a prediction parameter read from the storage device, and outputting the predicted traffic information via the communication device; It is characterized by having.

  In addition, on general roads, the traffic condition prediction device 20 takes in the current and past traffic information from the outside via a communication device, and statistical traffic information obtained by statistically processing the past traffic information. A step of calculating, a step of calculating a divergence degree quantitatively indicating the similarity between the current and past traffic information and the statistical traffic information, and a future traffic by the arithmetic unit using the divergence degree and the statistical traffic information. And a step of predicting and calculating the situation and outputting the predicted traffic information generated as a result via the communication device.

  As a result, according to the present invention, it is possible to select an appropriate prediction model according to the road type, and in particular, to predict traffic conditions with high accuracy continuously regardless of the presence or absence of sudden events on expressways. it can. Further, by calculating the traffic jam trend based on the predicted traffic information and using it together with the current traffic jam information, it becomes possible to make the information more detailed than in the past. Furthermore, according to the present invention, by providing the predicted traffic information to a device in which an application such as route search or dispatch and delivery plan for a transport vehicle operates, the user can more reliably reach the destination in the shortest time. You will be able to get

  The traffic situation prediction apparatus of the present invention is not only realized by dedicated hardware, but also records a program for realizing the function on a computer-readable recording medium, and the program recorded on the recording medium It may be read by a computer system and executed. Note that the computer-readable recording medium dynamically holds the program as in the case where the program is transferred via a communication medium such as the Internet in addition to a recording medium such as a flexible disk, a hard disk, and an optical disk. Or, it is assumed to include a program that holds a program for a certain period of time, such as a volatile memory of a server at that time.

  The present invention can be similarly applied to a traffic information center that predicts a future traffic situation using traffic information collected up to now or an in-vehicle device such as a car navigation system.

It is a figure which shows schematic structure of the traffic information system containing the traffic condition prediction apparatus in connection with this invention embodiment. It is a figure which shows an example of the delivery format of the prediction traffic information produced | generated by the traffic condition prediction apparatus concerning this invention embodiment. It is the block diagram which expanded and showed the function of the internal structure of the traffic condition prediction part among the traffic condition prediction apparatuses concerning this invention embodiment. It is a hardware block diagram which shows the internal structure of a vehicle-mounted apparatus among the traffic information systems shown in FIG. It is a flowchart which shows the process sequence of the route search process using the prediction traffic information of the traffic condition prediction apparatus concerning this invention embodiment. It is a figure which shows an example of the data structure of the heap table used for the route search process using the predicted traffic information of the traffic condition prediction apparatus concerning this invention embodiment. It is a flowchart which shows the process sequence of the whole traffic condition prediction process of the traffic condition prediction apparatus concerning this invention embodiment. It is the flowchart quoted in order to demonstrate the procedure of the prediction link information generation process for highways of the traffic condition prediction apparatus concerning this invention embodiment. It is the flowchart quoted in order to demonstrate the procedure of the prediction parameter generation process for highways of the traffic condition prediction apparatus concerning this invention embodiment. It is the flowchart quoted in order to demonstrate the procedure of the prediction process for highways of the traffic condition prediction apparatus concerning this invention embodiment. It is the flowchart quoted in order to demonstrate the procedure of the prediction process for general roads of the traffic condition prediction apparatus concerning this invention embodiment. It is a figure which shows an example of the data format of the prediction parameter for highways produced | generated by the traffic condition prediction apparatus concerning this invention embodiment. It is an example of the correction coefficient for correct | amending the estimated travel time used in the traffic condition prediction apparatus concerning this invention embodiment. It is an example of the correction coefficient for correct | amending a deviation degree used in the traffic condition prediction apparatus in connection with this invention embodiment in the situation where a deviation degree is small. It is an example of the correction coefficient for correcting a statistical travel speed in the situation where the degree of divergence is small, which is used in the traffic situation prediction apparatus according to the embodiment of the present invention. It is an example of the correction coefficient for correct | amending a statistical travel speed in the situation of a large divergence degree used in the traffic condition prediction apparatus concerning this invention embodiment. It is an example of the conditional expression for determining traffic congestion tendency information from the travel time change rate used in the traffic condition prediction apparatus according to the embodiment of the present invention. It is an example in the case of displaying on the display of a vehicle-mounted apparatus using the traffic congestion tendency information and the present condition traffic information which are produced | generated by the traffic condition prediction apparatus concerning this invention embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Network 11 Wireless base station 12 Vehicle 13 In-vehicle apparatus 14 Traffic information center (computer)
20 Traffic Situation Prediction Device 21 Data / Request Receiving Unit 22 Information Receiving / Storage Unit 23 Traffic Situation Prediction Unit 24 Prediction Information Distribution Unit 25 Route Search Unit 30 Storage DB
31 Statistics DB
32 Prediction DB

Claims (10)

At least a traffic situation prediction method for predicting a future traffic situation using current and past traffic information by a traffic situation prediction device configured by a storage device, a computing device, and a communication device,
The traffic situation prediction apparatus
A first step of capturing current and past traffic information from the outside via the communication device;
It is a prediction parameter that constitutes a prediction model using the propagation phenomenon of traffic conditions, and one or more propagation source links related to each prediction target link are associated, and each propagation source link includes a propagation time and a weighting factor. A second step of generating a prediction parameter constituting a prediction model using a propagation phenomenon of traffic conditions and storing it in the storage device;
Based on the prediction model using the traffic information and a prediction parameter read from the storage device, the arithmetic device predicts and calculates a future traffic situation, and outputs the predicted traffic situation via the communication device. And the steps
It was carried out,
The second step includes
A link that is n-th connected to the prediction target link is searched from map data including link attribute information, and among the searched links, a link within a predetermined distance from the prediction target link, the link length is longer than a predetermined length. Including a step of setting a link corresponding to at least one of a link, a link whose link type is a main line, and a link for which traffic information is provided as the propagation source link,
Traffic situation prediction how to, characterized in that.
The link order n in the n-order connection is defined as a link directly connected to the upstream side of the prediction target link as a primary link and a link directly connected to the upstream side of the primary link as a secondary link. This is the order when the link is directly connected to the upstream side of the secondary link as a tertiary link (n ≧ 1). The second step includes
Calculate the similarity between the traffic information of the prediction target link and the traffic information obtained by shifting the traffic information of the propagation source link in the past by the propagation time prepared for each, and the propagation time with the highest similarity is calculated. Determined as the propagation time of the propagation source link, using the traffic information of the prediction target link and the traffic information of the propagation source link shifted in the past based on the selected propagation time, the error of each traffic information Calculating the weighting factor that minimizes
The traffic condition prediction method according to claim 1, wherein: The third step includes
Among the one or more propagation source links, when it is determined that the proportion of the links whose travel speed is determined to be higher than the predetermined free running speed exceeds the predetermined value, the current traffic information is predicted as traffic conditions. Including the steps
The traffic condition prediction method according to claim 1 or claim 2 , wherein The third step includes
Of the one or more propagation source links, when it is determined that the proportion of links determined to have a travel speed higher than a predetermined free running speed is equal to or less than a predetermined value, the traffic information and the prediction parameter using the door, seeking expected travel speed in the prediction target time indicating the time width until the prediction target time from the present, to calculate the expected travel time on the link length included in the prediction target link is divided by the expected travel speed Steps,
Each travel time calculated by each of the weighting factor included in the prediction parameter, the statistical information obtained by statistically processing the past traffic information, and each weighting factor of the current traffic information, which is determined according to the prediction destination time. And correcting the estimated travel time for each predicted destination time,
The traffic condition prediction method according to claim 1 or 2 , characterized by comprising: Using current traffic information and the predicted traffic situation, the degree of traffic congestion at each prediction destination time indicating the time width from the current time to the prediction target time is quantified as a change rate compared to the current time, and the change rate and A traffic jam trend information generation step for selecting and outputting traffic jam trend information defined in advance by comparing with a threshold;
The traffic condition prediction method according to any one of claims 1 to 4 , further comprising:   Comparing the calculated similarity with a preset similarity determination threshold and excluding a propagation link having a similarity lower than the similarity determination threshold from a propagation source link used for prediction
The traffic condition prediction method according to claim 2, wherein:   For a plurality of propagation source links, calculate a cross-correlation coefficient between the propagation source links,
  When the cross-correlation coefficient value is compared with a cross-correlation coefficient determination threshold, and when a source link pair whose cross-correlation coefficient exceeds the cross-correlation coefficient determination threshold is found, Excluding the source link from the source link used for prediction
The traffic condition prediction method according to claim 2 or claim 6, wherein: A traffic situation prediction device that predicts future traffic situations using current and past traffic information,
A traffic information acquisition unit that captures the current and past traffic information;
One or more related transmission source links are associated with each prediction target link, and a prediction parameter that constitutes a prediction model using a propagation phenomenon of a traffic situation, which includes a propagation time and a weighting coefficient, is generated for each propagation source link. A prediction parameter generation unit to perform,
Predicting and calculating a future traffic situation based on the prediction model using the traffic information and the prediction parameter, and outputting a predicted traffic situation generation unit,
Equipped with a,
The prediction parameter generation unit
A link that is n-th connected to the prediction target link is searched from map data including link attribute information, and among the searched links, a link within a predetermined distance from the prediction target link, the link length is longer than a predetermined length. A link corresponding to at least one of a link, a link whose link type is a main line, and a link for which traffic information is provided is set as the propagation source link.
Traffic situation prediction apparatus according to claim and this.
The link order n in the n-order connection is defined as a link directly connected to the upstream side of the prediction target link as a primary link and a link directly connected to the upstream side of the primary link as a secondary link. This is the order when the link is directly connected to the upstream side of the secondary link as a tertiary link (n ≧ 1). Using current traffic information and the predicted traffic situation, the degree of traffic congestion at each prediction destination time indicating the time width from the current time to the prediction target time is quantified as a change rate compared to the current time, and the change rate and A traffic jam information generation / output unit that selectively outputs traffic jam trend information defined in advance by comparing with a threshold value,
The traffic condition prediction apparatus according to claim 8 , further comprising: The traffic situation prediction method according to any one of claims 1 to 7, a program for causing a computer to execute.

Images