JP6685763B2 - Accident prediction system and accident prediction method - Google Patents

Description

  Embodiments of the present invention relate to an accident prediction system and an accident prediction method.

  Conventionally, learning is performed using traffic data measured in the past at each point on the road (including traffic data measured when an accident occurs), and accidents (traffic accidents) at each point are likely to occur. A technique for forecasting (predicting) is known.

JP, 2014-35639, A JP, 2010-39992, A

  In the above-described conventional techniques, it is desired to predict an accident on a road with higher accuracy (prediction) and to reduce the calculation load.

The accident prediction system in the embodiment includes an accident prediction table creation processing unit, a traffic data acquisition processing unit, and an accident prediction processing unit. The accident prediction table creation processing unit uses past traffic data and past accident data for each of a plurality of roads divided into a plurality of sections, for each section group into which the sections are classified based on road conditions. Then, an accident occurrence pattern is learned based on a predetermined learning algorithm, and an accident prediction table indicating the likelihood of occurrence of an accident for each traffic situation is created. The traffic data acquisition processing unit acquires current traffic data for each section from a sensor that measures traffic conditions. The accident prediction processing unit, for each of the plurality of roads, for each section, the current traffic data of the section, and using the accident prediction table corresponding to the section group into which the section is classified , Predict the likelihood of occurrence.

FIG. 1 is a block diagram showing an example of the configuration of the accident prediction system in the first embodiment. FIG. 2 is an explanatory diagram that schematically shows an example of the relationship between the road sensor unit and the section that is the unit of the accident forecast regarding the road in the first embodiment. FIG. 3 is a flowchart showing an example of the flow of processing executed by the accident prediction table creation processing unit in the first embodiment. FIG. 4 is a diagram showing an example of a general configuration of the self-organizing map used in the first embodiment. FIG. 5 is a diagram showing an example of a more specific configuration of the self-organizing map used in the first embodiment. FIG. 6A is a diagram showing an example of a more specific configuration of the self-organizing map used in the first embodiment. FIG. 6B is a diagram showing an accident forecast table corresponding to the self-organizing map of FIG. 6A. FIG. 7 is a flowchart showing an example of the flow of processing executed by the accident prediction processing unit in the first embodiment. FIG. 8 is an explanatory diagram when an accident forecast is performed using the self-organizing map of FIG. FIG. 9 is an explanatory diagram of a specific example 1 in the first embodiment. FIG. 10 is a diagram showing an example of the configuration of the accident forecast table database in the case of the specific example 1 in the first embodiment. FIG. 11 is a flowchart showing an example of the flow of processing executed by the accident prediction processing unit in the case of the first specific example of the first embodiment. FIG. 12 is an explanatory diagram of a specific example 2 in the first embodiment. FIG. 13 is a diagram showing an example of the configuration of the accident forecast table database in the case of the second specific example of the first embodiment. FIG. 14 is a flowchart showing an example of the flow of processing executed by the accident prediction processing unit in the case of the second specific example of the first embodiment. FIG. 15 is a block diagram showing an example of the configuration of the accident prediction system according to the second embodiment. FIG. 16 is a block diagram showing an example of the configuration of the accident prediction system in the third embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following, "prediction" means calculating / predicting the risk of a traffic accident based on past traffic data at each point on the road (including traffic data at the time of the accident) It includes the meaning of notifying as a forecast, but also means not only notifying the prediction result but also simply “predicting” the likelihood of an accident at each point. Further, the term “route” includes the meaning of a route in the Road Law, but is not limited to this, and includes the meaning of a road as a management unit.

(First embodiment)
First, the configuration of the accident prediction system 1 according to the first embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 is a block diagram showing an example of the configuration of an accident prediction system 1 according to the first embodiment. FIG. 2 is an explanatory diagram that schematically shows an example of the relationship between the road sensor unit and the section that is the unit of the accident forecast regarding the road in the first embodiment. The white arrow Y in FIG. 2 indicates the flow direction of the vehicle on the road R (hereinafter, the description of “R” may be omitted).

The accident prediction system 1 shown in FIG. 1 is measured by the road sensor units RS ₁ , RS ₂ , RS ₃ , ... Corresponding to the sections 1, 2, 3, ... On the road R shown in FIG. It is a system that predicts the likelihood of occurrence of traffic accidents (risk level of accidents) in each section using traffic data. In addition, in the following description using FIG. 2, it is assumed that the sections are three sections 1 to 3 in order to simplify the description.

The road sensor units RS _{1 to} RS ₃ each include a sensor (sensing device) capable of detecting a vehicle traveling in the sections 1 to 3 on the road R. This sensor is composed of, for example, a loop coil installed below the road surface, a camera or an ultrasonic sensor that monitors the road surface from above. Hereinafter, when the road sensor units RS _{1 to} RS ₃ are not particularly distinguished, they are collectively referred to as the road sensor unit RS.

  In addition, the road sensor unit RS includes a traffic data processing unit. Specifically, the traffic data processing unit, based on the traffic data measured by the sensor, the traffic volume [vehicles / hour] of vehicles traveling on the road R, the average speed [km / h], the vehicle density [vehicles / vehicle]. km], occupancy (occupancy) [%], etc. are calculated, and the calculation result is transmitted to the road traffic control device 2. This calculation and transmission are executed in time units such as 1 minute and 5 minutes. The road sensor unit RS may transmit only the measurement result by the sensor to the traffic data acquisition processing unit 211 of the road traffic control device 2, and the traffic data acquisition processing unit 211 may calculate the traffic volume and the like.

  Here, in the first embodiment, the accident prediction system 1 includes a road traffic control device 2 and an accident prediction table creation device 3. The road traffic control device 2 is, for example, a computer system generally called a road traffic control system. Although the road traffic control device 2 is shown as one computer device in FIG. 1 for the sake of simplicity of description, it may be realized by a plurality of computer devices.

  The road traffic control device 2 includes a processing unit 21, a storage unit 22, a display unit 23, and an input unit 24. The road traffic control device 2 also has a communication unit for communication with an external device, but illustration and description thereof are omitted for simplicity of description.

  The processing unit 21 controls the overall operation of the road traffic control device 2 and realizes various functions of the road traffic control device 2. The processing unit 21 includes, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). The CPU centrally controls the operation of the road traffic control device 2. The ROM is a storage medium that stores various programs and data. The RAM is a storage medium for temporarily storing various programs and rewriting various data. The CPU executes the programs stored in the ROM, the storage unit 22 and the like, using the RAM as a work area (work area). The processing unit 21 includes a traffic data acquisition processing unit 211, an accident forecast processing unit 212, a reception processing unit 213, and a display control unit 214.

  The traffic data acquisition processing unit 211 periodically acquires the measured traffic data from each road sensor unit RS installed for each section of the road. Then, the traffic data acquisition processing unit 211 transmits the acquired traffic data so as to be accumulated as current traffic data in the current database 221 of the storage unit 22, and also stores the past traffic data in the storage unit 32 of the accident prediction table creation device 3. Send to store as traffic data. When there are a plurality of lanes on the road and the road sensor unit RS is set corresponding to each lane, the traffic data acquisition processing unit 211 may integrate the traffic data of each lane, for example. Further, in the present embodiment, of traffic data, for example, the traffic data for the last few minutes is referred to as current traffic data, and the past long-term traffic data including the current traffic data is referred to as past traffic data. The past traffic data also includes traffic data measured when an accident occurred.

  The accident prediction processing unit 212 uses, for each section of the road, current traffic data (for example, traffic volume, average speed, vehicle density) of the section and an accident prediction table (details will be described later) corresponding to the section. And predict the likelihood of an accident.

  The reception processing unit 213 stores the accident prediction table (details will be described later) received from the transmission processing unit 312 of the accident prediction table creation device 3 in the accident prediction table database 222 of the storage unit 22. When there are a plurality of accident prediction tables, the reception processing unit 213 stores them in the accident prediction table database 222 together with their identification information.

  The display control unit 214 controls the display unit 23 to display the accident prediction result (the accident occurrence degree described later) by the accident prediction processing unit 212 as an accident prediction. For example, it is preferable to notify the control personnel by displaying the alarm lamp of the display unit 23 in a lighting manner, for example, in a section where the degree of accident occurrence is high, because an accident is likely to occur. In the road traffic control device 2, when the warning lamp is lighted and displayed as described above, for example, an alarm sound may be sounded by a voice output means (not shown).

  The storage unit 22 is a storage device such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). The storage unit 22 stores a current database 221 and an accident forecast table database 222.

  The current database 221 stores the most recent traffic data (current traffic data) acquired by the traffic data acquisition processing unit 211, for example, for a few minutes. Further, in the current database 221, the traffic data acquired by the traffic data acquisition processing unit 211 is accumulated, and at the same time, road characteristic data representing the characteristics of the target road (for example, road length, sensor installation position, peripheral information of each measurement point, (Toll gate location, etc.), policy information, accident information, construction information, and other information managed in road traffic control, speed limit information, etc. are stored. The road characteristic data representing the characteristic of the target road may be input at the time of system construction in advance, but may be additionally corrected by a controller or the like. Information managed in road traffic control such as measure information, accident information, construction information, and speed limit information may be manually input by a user (a controller, etc.) of the road traffic control device 2, for example. Then, the current traffic data in the current database 221 is used when the accident prediction processing unit 212 makes an accident prediction (prediction). At this time, when carrying out an accident forecast (prediction), the road characteristic data currently stored in the database 221 is referred to, and the current traffic data corresponding to the relevant part of the accident forecast is set in the data set (traffic volume [vehicle / h ], Average speed [km / h], vehicle density [vehicles / km] (or set of occupancy [%])).

  The accident prediction table database 222 stores one or more accident prediction tables (details will be described later).

  The display unit 23 displays, for example, the forecast result of the accident by the accident forecast processing unit 212. The display unit 23 is realized by, for example, a liquid crystal display device (LCD (Liquid Crystal Display)), an organic EL (Electro-Luminescence) display device, or the like.

  The input unit 24 receives a user operation on the road traffic control device 2. The input unit 24 is, for example, an input device such as a keyboard and a mouse.

  The accident forecast table creating device 3 is a computer device for creating an accident forecast table. The accident forecast table creation device 3 includes a processing unit 31, a storage unit 32, a display unit 33, and an input unit 34. The accident prediction table creation device 3 also has a communication unit for communication with an external device, but illustration and description thereof are omitted for simplicity of description.

  The processing unit 31 controls the entire operation of the accident prediction table creation device 3 and realizes various functions of the accident prediction table creation device 3. The processing unit 31 includes, for example, a CPU, a ROM, and a RAM. The CPU centrally controls the operation of the accident prediction table creation device 3. The ROM is a storage medium that stores various programs and data. The RAM is a storage medium for temporarily storing various programs and rewriting various data. The CPU executes the programs stored in the ROM, the storage unit 32, etc., using the RAM as a work area (work area). The processing unit 31 includes an accident forecast table creation processing unit 311 and a transmission processing unit 312.

  The accident prediction table creation processing unit 311 collects past traffic data and past accident data (details will be described later) in the storage unit 32 for each predetermined set of sections regarding a plurality of roads that are divided into a plurality of sections. Accident occurrence for each traffic situation (traffic volume, average speed, vehicle density, occupancy rate, etc.) by learning an accident occurrence pattern based on a predetermined learning algorithm (for example, a learning algorithm using a self-organizing map) Create an accident forecast table that indicates how easy it is (details will be described later). Here, a self-organizing map is a type of neural network that is applied to important fields in the real world such as process analysis, control, search system, and information analysis for management, and it is a high-dimensional input data. Is an algorithm for clustering without a priori knowledge such as a teacher signal (output considered to be ideal for input data). The specific contents of this self-organizing map will be described later.

  Further, the accident prediction table creation processing unit 311 uses the past traffic data and the past accident data acquired at that time at a predetermined timing or when an instruction is input by the user, to use the accident prediction table. To update. The predetermined timing is, for example, every year, when the past traffic data and past accident data for the most recent year are accumulated. At this time, an accident forecast table is created using the data for one year accumulated in the accident prediction table creation device 3, and the accident prediction table is transmitted to the road traffic control device 2 for accident prediction. The accident forecast table of the table database 222 is updated. Also, when there is an instruction input by the user, for example, when the flow of vehicles on the target road changes due to the formation of a large road around the target road, etc. This is a case where the user inputs an instruction for updating the accident prediction table using the input unit 34 of the accident prediction table creation device 3 when a sufficient amount of past traffic data and past accident data are accumulated. .

  The transmission processing unit 312 sends the accident prediction table (details described later) created (created for the first time and updated) by the accident prediction table creation processing unit 311 to the reception processing unit 213 of the processing unit 21 of the road traffic control device 2. Send.

  The storage unit 32 is a storage device such as an HDD or SSD. The storage unit 32 stores the past database 321. The past database 321 stores past traffic data and past accident data. The past traffic data is sequentially accumulated by the traffic data received from the traffic data acquisition processing unit 211 of the road traffic control device 2.

  The past accident data is data of past accidents that occurred on the target road. The past accident data may be stored in the past database 321 of the storage unit 32, for example, by the user inputting the past accident data using the input unit 34 of the accident prediction table creation device 3 while looking at the accident report or the like. . The past accident data specifically includes, for example, an accident occurrence point, an accident occurrence date and time, an accident type, and the like as information about the accident. The past accident data is preferably accident information for the past several years or more. Further, the past accident data is input by the user at the input unit 34 of the accident forecast table creation device 3, and at the same time by the user at the input unit 24 of the road traffic control device 2 from the road traffic control device 2 to the accident forecast table. The data may be stored in the past database 321 of the storage unit 32 by transmitting it to the creation device 3. Alternatively, past accident data in another computer device is stored in the past database 321 of the storage unit 32 of the accident prediction table creating device 3 via an information storage medium such as a DVD (Digital Versatile Disk) or a USB (Universal Serial Bus) memory. It may be stored in.

  The display unit 33 displays various screens. The display unit 33 is realized by, for example, a liquid crystal display device (LCD), an organic EL display device, or the like.

  The input unit 34 receives a user's operation on the accident prediction table creation device 3. The input unit 34 is, for example, an input device such as a keyboard and a mouse.

  Further, for example, the accident prediction table creation processing unit 311 of the accident prediction table creation device 3 creates an accident prediction table for each road as a predetermined set of sections. At that time, the accident prediction processing unit 212 of the road traffic control device 2 indicates, for each of the plurality of roads, current traffic data (for example, traffic volume, average speed, vehicle density) of the section and the road. The likelihood of an accident is predicted (predicted) using the corresponding accident prediction table (described later with FIGS. 9 to 11).

  Further, for example, the accident prediction table creation processing unit 311 of the accident prediction table creation device 3 sets, for each predetermined set of sections, for each of a plurality of roads, for each section group in which the sections are classified based on road conditions. , Create an accident forecast table. At that time, the accident prediction processing unit 212 of the road traffic control device 2 determines, for each of the plurality of roads, current traffic data (for example, traffic volume, average speed, vehicle density) of the section and the section. The likelihood of an accident is predicted (predicted) using the accident prediction table corresponding to the classified section groups (described later with FIGS. 12 to 14).

  Next, the operation of the accident prediction system 1 according to the first embodiment will be described. First, the operation of the accident prediction table creation processing unit 311 in the first embodiment will be described with reference to FIG. FIG. 3 is a flowchart showing an example of the flow of processing executed by the accident prediction table creation processing unit 311 in the first embodiment.

  As shown in FIG. 3, the accident prediction table creation processing unit 311 first reads past traffic data and past accident data from the past database 321 in step S1.

  Next, in step S2, the accident prediction table creation processing unit 311 learns the correlation between the traffic data and the probability of occurrence of an accident based on the past traffic data and the past accident data read in step S1. As a learning method, for example, a method using the following self-organizing map is used.

FIG. 4 is a diagram showing an example of a general configuration of the self-organizing map used in the first embodiment. As shown in FIG. 4, the self-organizing map is a two-layer neural network having an input layer and a competitive layer (output layer). The input layer is represented as a plane having the same number of units as the data x ₁ , ..., X _i , ..., X _n to be analyzed. Here, the combination of the data x ₁ , ..., X _i , ..., X _{n to be} analyzed is called an input vector. Further, the competitive layer is represented as a plane including a plurality of units 1, ..., J ,. Each unit of the input layer and each unit of the competitive layer are related by a weight vector w _j = (w _j1 , ..., W _ji , ..., W _jn ) having the same dimension as the input vector.

  FIG. 5 is a diagram showing an example of a more specific configuration of the self-organizing map used in the first embodiment. As shown in FIG. 5, the past traffic data read from the past database 321 is used as an input vector to be learned. For example, the past traffic data is composed of three types of data (traffic volume, average speed, and vehicle density) measured at a certain point in the past at each point on the road R.

  In the self-organizing map of FIG. 5, first, the process of (1) determining the winner unit and updating the weight vector of the winner unit is executed. Then, (2) a process of updating the weight vector of a neighboring unit located near the winner unit is executed. The winner unit is one unit on the competitive layer that is associated with the input vector by the weight vector most similar to the input vector. Further, the updating of the weight vector is performed by using a mathematical formula that considers the number of times of learning and a predetermined learning coefficient. The details of the method of determining the weight vector most similar to the input vector and the details of the mathematical formula used for updating the weight vector are disclosed in, for example, Japanese Patent Application Laid-Open No. 2014-35639, and therefore, no further explanation will be given here. Omit it.

  Further, the accident degree set corresponding to the past traffic data used as the input vector is input. The degree of accident occurrence is a value indicating the likelihood of occurrence of an accident, which is set corresponding to the input vector. When each element of the input vector is traffic data when an accident occurs, the accident occurrence degree is set to, for example, "1" (or "100"). When each element of the input vector is traffic data when there is no accident, the accident occurrence rate is set to “0”, for example. Further, when each element of the input vector is traffic data immediately before the occurrence of the accident, the accident occurrence rate is set to a value smaller than the accident occurrence rate (“1” (or “100”)) at the time of the accident occurrence.

  In the first embodiment, after the processes of (1) and (2) described above are executed, (3) the accident occurrence distribution corresponding to the winning unit and the neighboring unit on the competitive layer with the inputted accident occurrence degree. The process of updating the value of the unit on the layer is executed. Specifically, the value of the unit corresponding to the winner unit is updated by the inputted accident occurrence degree, and the value of the unit corresponding to the neighboring unit is updated by the value smaller than the inputted accident occurrence degree. The details of this updating process are also disclosed in the above-mentioned Japanese Patent Laid-Open No. 2014-35639, and therefore further description is omitted here.

  In the first embodiment, the correlation between the past traffic data and the accident occurrence degree is learned by repeatedly executing the above processes (1) to (3).

  Returning to FIG. 3, in step S3, the accident prediction table creation processing unit 311 creates an accident prediction table based on the learning result in step S2. That is, the accident prediction table creation processing unit 311 creates a table for the correlation obtained by the self-organizing map shown in FIG. 5, and creates an accident prediction table that defines the correlation between the past traffic data and the accident occurrence rate. create. A more specific configuration example of the self-organizing map and an example of the accident forecast table corresponding to the self-organizing map will be described with reference to FIG.

  FIG. 6A is a diagram showing an example of a more specific configuration of the self-organizing map used in the first embodiment. FIG. 6B is a diagram showing an accident forecast table corresponding to the self-organizing map of FIG. 6A.

  As shown in FIG. 6A, traffic volume, average speed, and vehicle density data are input as input vectors to the input layer. Here, it is assumed that there are 100 (10 × 10) units 1-1, 1-2, ..., 10-10 in the competitive layer. Moreover, the range of the accident occurrence degree is set to "0" to "10". In this case, the accident forecast table is as shown in FIG. In the accident prediction table shown in FIG. 6B, for each index (corresponding to a competitive layer or a unit of accident occurrence distribution), a weight for traffic volume, a weight for average speed, a weight for vehicle density, and an accident occurrence rate. Are associated with. How to use the accident forecast table of FIG. 6B will be described later.

  Next, the operation of the accident forecast processing unit 212 in the first embodiment will be described with reference to FIG. 7. FIG. 7 is a flowchart showing an example of the flow of processing executed by the accident forecast processing unit 212 in the first embodiment. Note that the outline of the process of the accident forecast processing unit 212 will be described with reference to FIG. 7, and a specific example of the process of the accident forecast processing unit 212 will be described with reference to FIGS. 11 and 14.

  As shown in FIG. 7, the accident prediction processing unit 212 first determines in step S11 whether or not the prediction timing has arrived. If Yes, the process proceeds to step S12, and if No, the process returns to step S11. The forecast timing may be, for example, every 5 minutes, but is not limited to this.

  Next, the accident forecast processing unit 212 performs processing for each route in steps S12 to S17. That is, when the accident forecast (prediction) for each section is performed for a plurality of routes, the accident forecast processing unit 212 first performs the processes of steps S13 to S16 for the first route and then the steps for the second route. The processes of S13 to S16 are performed, and the process of ... Is sequentially performed for all routes.

  As described above, the accident forecast processing unit 212 performs processing for each section in steps S13 to S16. That is, the accident forecast processing unit 212 first performs the processes of steps S14 and S15 for the first section, then the steps S14 and S15 of the second section, for the route of interest, ... , Is performed for all sections.

  In step S14, the accident prediction processing unit 212 reads current traffic data (traffic volume, average speed, vehicle density in the example of FIG. 6) from the current database 221 of the storage unit 22 for the section of interest.

  Next, in step S15, the accident prediction processing unit 212 reads an accident prediction table corresponding to the section from the accident prediction table database 222 of the storage unit 22 and uses it to perform an accident prediction (prediction).

  Here, step S15 will be described with reference to FIG. FIG. 8 is an explanatory diagram when an accident forecast (prediction) is performed using the self-organizing map of FIG. As shown in FIG. 8, the following processes (11) to (13) are executed in this self-organizing map.

(11) The winner unit is determined (for example, “2-1” of the competitive layer in FIG. 6A). The method of determining the winner unit is the same as that in the case of (1) (FIG. 5) described above.
(12) Select a unit on the layer of the accident occurrence distribution corresponding to the winner unit (for example, “2-1” of the accident occurrence distribution in FIG. 6A).
(13) Output the accident occurrence degree of the selected unit (for example, the accident occurrence degree “6.2” corresponding to index “2-1” in the accident prediction table of FIG. 6B).

  In this way, the accident prediction processing unit 212 can acquire the accident occurrence degree of each section for each of a plurality of routes in steps S12 to S17.

  In step S18, the display control unit 214 controls to display the accident prediction result (accident occurrence degree, etc.) by the accident prediction processing unit 212 on the display unit 23 together with an alarm lamp as an accident prediction. As a result, a controller or the like using the road traffic control device 2 can recognize the likelihood of an accident corresponding to the current traffic situation for each section of the target road. Moreover, since the accident prediction table is created and used for each predetermined set of sections, it is possible to predict (predict) an accident on the road with higher accuracy and to reduce the calculation load.

(Specific example 1)
Next, a specific example 1 of the first embodiment will be described with reference to FIGS. 9 to 11. FIG. 9 is an explanatory diagram of the first specific example. In this specific example 1, the target lines (roads) are two lines A and B. Then, one accident forecast table for route A is used for all sections, and another accident forecast table for route B is used for all sections. Hereinafter, the accident prediction table (self-organizing map) will be referred to as SOM (Self-Organizing Map).

As shown in FIG. 9A, the route A is divided into n sections (sections 1 to n). Road sensor units RS (RS _{A1 to} RS _An ) are provided in each section. Each road sensor unit RS outputs traffic volume Q, average speed V, and vehicle density D as current traffic data. For example, the road sensor unit RS _A1 outputs the traffic volume Q _A1 , the average speed V _A1 , and the vehicle density D _A1 . The output current traffic data (traffic volume Q, average speed V, vehicle density D) is stored in the current database 221 (FIG. 1) of the storage unit 22. In addition, the common SOM _A is used for all sections of the route A. SOMA _A is created using past traffic data and past accident data for all sections of route A.

Similarly, as shown in FIG. 9B, the route B is divided into m sections (sections 1 to m). Road sensor units RS (RS _{B1 to} RS _Bm ) are provided in each section. Each road sensor unit RS outputs traffic volume Q, average speed V, and vehicle density D as current traffic data. For example, the road sensor unit RS _B1 outputs the traffic volume Q _B1 , the average speed V _B1 , and the vehicle density D _B1 . The output current traffic data (traffic volume Q, average speed V, vehicle density D) is stored in the current database 221 (FIG. 1) of the storage unit 22. In addition, the common SOM _B is used for all sections of the route B. SOM _B is created using past traffic data and past accident data for all sections of route B.

Here, FIG. 10 is a diagram showing an example of the configuration of the accident forecast table database 222 in the case of the first specific example. As shown in FIG. 10, SOM _A and SOM _B are stored in the accident prediction table database 222.

  Next, with reference to FIG. 11, the processing executed by the accident forecast processing unit 212 in the case of the specific example 1 will be described. FIG. 11 is a flowchart showing an example of the flow of processing executed by the accident prediction processing unit 212 in the case of the first specific example.

  As shown in FIG. 11, the accident prediction processing unit 212 first determines in step S201 whether or not the prediction timing has arrived. If Yes, the process proceeds to step S202, and if No, the process returns to step S201. The forecast timing may be, for example, every 5 minutes.

Next, the accident forecast processing unit 212 performs processing for each section of the route A in steps S202 to S205. Specifically, in step S203, the accident prediction processing unit 212 uses the current traffic data (traffic volume Q _A1 , average speed V _A1 , vehicle density D _A1 ) from the current database 221 of the storage unit 22 for the section 1 of the route A. Read out. Next, in step S204, the accident forecast processing unit 212 reads SOM _A from the accident forecast table database 222 of the storage unit 22, and obtains the current traffic data (traffic volume Q _A1 , average speed V _A1 , vehicle density D _A1 ). An accident forecast (prediction) is performed using SOM _A , and a forecast result RE _A1 (FIG. 9A) is obtained. The accident prediction processing unit 212 similarly performs the accident prediction (prediction) for the sections 2 to n of the route A, and obtains the prediction results RE _{A2 to} RE _An (FIG. 9A).

Next, the accident forecast processing unit 212 performs processing for each section of the route B in steps S206 to S209. Specifically, in step S207, the accident prediction processing unit 212 uses the current traffic data (traffic volume _QB1 , average speed _VB1 , vehicle density _DB1 ) from the current database 221 of the storage unit 22 for the section 1 of the route B. Read out. Next, in step S208, the accident prediction processing unit 212 reads SOM _B from the accident prediction table database 222 of the storage unit 22 and uses it as current traffic data (traffic volume Q _B1 , average speed V _B1 , vehicle density D _B1 ). An accident forecast (prediction) is performed using SOM _B , and a forecast result RE _B1 (FIG. 9B) is obtained. The accident prediction processing unit 212 similarly performs an accident prediction (prediction) for the sections 2 to m of the route B and obtains the prediction results RE _{B2 to} RE _Bm (FIG. 9 (b)).

  In step S210, the display control unit 214 performs control such that the forecast result (accident occurrence degree, etc.) of each section of the routes A and B is displayed as an accident forecast on the display unit 23 together with an alarm lamp and the like.

In this way, using a single SOM _A for route A, also, for the route B to use a single SOM _B, forecasts with higher accuracy accidents in light of the accident characteristics of each line In addition to (predicting), the calculation load (memory resource, calculation resource) can be significantly reduced as compared to the case where different SOMs are used for each section. In other words, in the example of FIG. 6, the number of units of the competitive layer in the SOM is 100 (10 × 10), but in many cases it is actually larger than that, and in that case, different SOMs are all used for each section. Then, the calculation load (memory resource, calculation resource) becomes excessive. On the other hand, according to the method of the present embodiment, it is possible to save memory resources and reduce computational resources by simplifying program processing.

(Specific example 2)
Next, the second specific example will be described with reference to FIGS. 12 to 14. As described above, it may be unfavorable in terms of calculation load to use different SOMs for all road sections. However, for one road, it may be effective to classify a plurality of sections into some section groups based on, for example, road conditions, and use one SOM in association with each section group. . This is because, according to that method, the increase in calculation load is not so large, and the characteristics of the likelihood of accidents due to road shapes and the like can be taken into consideration.

  The section group includes, for example, a group of sections having an uphill (uphill section group), a group of sections having a downhill (downhill section group), a group of sections having a curve (curve section group), and a branch. Group of sections (branch section group), group of sections with confluence (merging section group), group of sections with toll gates (toll gate section group), group of sections with tunnels (tunnel section group), with bridge Group of sections (bridge section group), group of sections with lane number increase points (lane number increase section group), group of sections with lane number decrease points (lane number decrease section group), group of other sections (normal) Section group) is possible.

  For example, an uphill, a downhill, and a curve are scenes in which the driver is forced to operate the accelerator, brake, and handle, and it is considered that the characteristics of the likelihood of accidents are affected.

  Moreover, it is considered that branching, merging, an increased number of lanes, and a decreased number of lanes have an influence on the characteristics of the likelihood of an accident in relation to other vehicles.

  Moreover, it is considered that the toll booth has a significant effect on the characteristics of the likelihood of accidents because it is necessary to temporarily reduce the speed significantly. Note that, for example, when the driving behavior of the driver is significantly different between the main toll gate and other toll gates, they may be treated separately.

  In addition, it is considered that the tunnel has an influence on the characteristics of the likelihood of an accident in that the brightness changes abruptly.

  In addition, the bridge is likely to freeze at low temperatures and has strong winds, which may affect the characteristics of the likelihood of accidents.

  In addition to the above, if you set a section group according to the road conditions that are likely to affect the characteristics of the likelihood of accidents, such as the northern side of the mountain where the temperature does not rise all day Good.

  Based on the above, specific example 2 will be specifically described below. FIG. 12 is an explanatory diagram of the second specific example. In this specific example 2, the target lines (roads) are two lines, line A and line B. Then, a different accident prediction table is used for each section group for the route A, and a different accident prediction table is used for each section group for the route B as well.

As shown in FIG. 12A, the route A is typically divided into 16 sections (sections 1 to 16). A road sensor unit RS is provided in each section, and each road sensor unit RS outputs traffic volume Q, average speed V, and vehicle density D as current traffic data, and the output current traffic data (traffic volume Q , The average speed V, and the vehicle density D) are stored in the current database 221 of the storage unit 22 as in the first specific example, and are not shown (the same applies to FIG. 12B). Further, sections 1 to 16 are classified into 6 types of section groups. Sections 1, 3, 5, 6, 8, 10, 11, 13, 15 are normal section groups. Sections 2 and 12 are merging section groups. Sections 4 and 14 are branching section groups. Section 7 is a downhill section group. Section 9 is an uphill section group. Section 16 is a tollgate section group. For each section group, the SOMs used are SOM _Aa , SOM _Ab , SOM _Ac , SOM _Ad , SOM _Ae , SOM _Af , in that order. Each SOM is created using the past traffic data and the past accident data of the section belonging to the section group.

Similarly, as shown in FIG. 12B, the route B is typically divided into 17 sections (sections 1 to 17). Further, the sections 1 to 17 are classified into seven types of section groups. Sections 2, 4, 6, 8, 10, 11, 12, 14, 16 are normal section groups. Sections 9 and 13 are merging section groups. Sections 7 and 15 are branching section groups. Section 3 is a downhill section group. Section 5 is an uphill section group. Section 1 is a tollgate section group. The section 17 is a curve section group. For each section group, the SOMs used are SOM _Ba , SOM _Bb , SOM _Bc , SOM _Bd , SOM _Be , SOM _Bf , SOM _Bg , in that order. Each SOM is created using the past traffic data and the past accident data of the section belonging to the section group.

Here, FIG. 13 is a diagram showing an example of the configuration of the accident forecast table database 222 in the case of the second specific example. As shown in FIG. 13, the accident forecast table database 222 stores SOM _Aa to SOM _Af for the route A and SOM _Ba to SOM _Bg for the route B.

  Next, with reference to FIG. 14, a process executed by the accident forecast processing unit 212 in the case of the specific example 2 will be described. FIG. 14 is a flowchart showing an example of the flow of processing executed by the accident forecast processing unit 212 in the case of the second specific example.

  As shown in FIG. 14, the accident forecast processing unit 212 first determines in step S301 whether or not the forecast timing has arrived. If Yes, the process proceeds to step S302, and if No, the process returns to step S301. The forecast timing may be, for example, every 5 minutes.

Next, the accident forecast processing unit 212 performs processing for each section of the route A in steps S302 to S305. Specifically, in step S303, the accident forecast processing unit 212 reads current traffic data (traffic volume, average speed, vehicle density) from the current database 221 of the storage unit 22 for the section 1 of the route A. Next, in step S304, the accident prediction processing unit 212 reads the SOM _Aa corresponding to the normal section group into which the section 1 is classified from the accident prediction table database 222 of the storage unit 22 and uses the current traffic data and the SOM _Aa . Accident forecast (prediction). In step S304, SOM _Ab is used in the section 2, SOM _Aa is used in the section 3, and SOM _Ac is used in the section 4. Also in the case of the sections 5 to 16, similarly, the SOM corresponding to the section group into which the section is classified is used.

Next, the accident forecast processing unit 212 performs processing for each section of the route B in steps S306 to S309. Specifically, in step S307, the accident prediction processing unit 212 reads the current traffic data (traffic volume, average speed, vehicle density) from the current database 221 of the storage unit 22 for the section 1 of the route B. Next, in step S308, the accident prediction processing unit 212 reads the SOM _Bf corresponding to the tollgate section group into which Section 1 is classified from the accident prediction table database 222 of the storage unit 22, and obtains the current traffic data and SOM _Bf . Use it to make an accident forecast. In step S308, SOM _Ba is used in the section 2, SOM _Bd is used in the section 3, and SOM _Ba is used in the section 4. Also in the case of the sections 5 to 17, similarly, the SOM corresponding to the section group into which the section is classified is used.

  In step S310, the display control unit 214 performs control such that the forecast result (accident occurrence degree, etc.) of each section of the routes A and B is displayed as an accident forecast on the display unit 23 together with an alarm lamp and the like.

  In this way, by using one SOM in association with each section group for one line, an accident can be predicted (predicted) with higher accuracy based on the accident characteristics of each section group, and The calculation load (memory resource, calculation resource) can be significantly reduced as compared to the case where different SOMs are used for each.

(Second embodiment)
Next, with reference to FIG. 15, an accident prediction system 1a of the second embodiment will be described. FIG. 15 is a block diagram showing an example of the configuration of the accident prediction system 1a according to the second embodiment. The difference between the accident prediction system 1a of FIG. 15 and the accident prediction system 1 of the first embodiment of FIG. 1 is that the road traffic control device 2a does not have a reception processing unit 213 and that the accident prediction table creation device 3a is different. The point is that there is no transmission processing unit 312. The description of the same points as those of the accident prediction system 1 of FIG. 1 will be omitted.

  In the accident prediction system 1 of the first embodiment of FIG. 1, the accident prediction table creation processing unit 311 of the processing unit 31 of the accident prediction table creation device 3 creates the accident prediction table for the first time and then at a predetermined timing. Or, when an instruction is input by the user, a new accident forecast table is created and updated using the past traffic data and past accident data acquired at that time. On the other hand, the accident forecast system 1a of FIG. 15 does not assume updating of the accident forecast table. After the accident prediction table creation processing unit 311 creates the accident prediction table for the first time, the accident prediction table is stored on the road traffic, for example, via the Internet or an information storage medium such as a DVD or a USB memory. The data is stored in the accident prediction table database 222 of the storage unit 22 of the control device 2a. After that, the accident forecast table is not updated.

  In this way, according to the accident prediction system 1a of the second embodiment, the structure and the processing are simplified because the accident prediction table is not updated. When it is not necessary to update the accident prediction table depending on the route, such an accident prediction system 1a of the second embodiment can be applied.

(Third Embodiment)
Next, with reference to FIG. 16, an accident prediction system 1b of the third embodiment will be described. FIG. 16 is a block diagram showing an example of the configuration of the accident prediction system 1b according to the third embodiment. The accident forecast system 1b of FIG. 16 is different from the accident forecast system 1 of the first embodiment of FIG. 1 in that the configuration of the road traffic control device 2 and the configuration of the accident forecast table creation device 3 are integrated into the accident forecast system. The point is that the system 1b is used. The accident prediction system 1b can be realized by using, for example, a conventional road traffic control system.

  Specifically, the accident prediction system 1b includes a processing unit 21a, a storage unit 22a, a display unit 23, and an input unit 24. The processing unit 21a includes a traffic data acquisition processing unit 211, an accident forecast processing unit 212, a display control unit 214, and an accident forecast table creation processing unit 311. The storage unit 22a stores a current database 221, an accident forecast table database 222, and a past database 321. The individual configuration (each part, each database) is the same as that of the first embodiment, and thus the description is omitted.

  In this way, according to the accident prediction system 1b of the third embodiment, since it is a single system, the configuration and processing are simple, and an accident prediction table using traffic data obtained in real time is used. Will be easier to update. In this case, for example, the accident forecast table may be automatically updated every day or every month when traffic data and accident data are accumulated, or an accident may occur each time a predetermined number of accident data is accumulated. A method of automatically updating the forecast table may be considered.

(Modification)
In the above embodiment, a method using a self-organizing map is illustrated as a method for performing learning and prediction (prediction). However, as a method of learning and prediction (prediction), various methods are conceivable other than the method using the self-organizing map. For example, as a relatively simple method, a method of holding (accumulating) past traffic data at the time of an accident and simply comparing it with the current traffic data, or a combination of past traffic data at the time of an accident is clustered by statistical processing, A method of generating traffic data for similar cases when an accident occurs can be considered. As another method, for example, a method using other multivariate analysis such as Paigean network can be considered.

  Further, in the above embodiment, the traffic volume, the average speed, the vehicle density, and the occupancy rate are exemplified as the data used for learning and prediction (prediction). However, even information other than these four types may be used for learning and forecasting (prediction) as long as the information has a correlation with the occurrence of an accident, and for example, the following (A) to (L ) Is available.

(A) Weather information (clear weather, rain, fog, snow, etc., temperature, humidity, etc.)
(B) Large vehicle mixture information (mixing rate of large vehicles such as trucks and buses)
(C) Low-speed vehicle mixing information (mixing rate of low-speed vehicles traveling at a speed below a predetermined value, etc.)
(D) Speed limit information (normal speed limit, temporary speed limit, etc.)
(E) Event information (event status information around the road, etc.)
(F) Hazard information (information about road conditions based on hazard maps, etc.)
(G) Motorcycle mixing information (mixing rate of motorcycles, etc.)
(H) Road maintenance information (road construction information, etc.)
(I) Closure information (information indicating the status of closure due to accidents, equipment troubles, etc.)
(J) Road surface condition information (information indicating road dryness, wetness, freezing, etc.)
(K) Congestion information (congestion length, start position, end position, etc.)
(L) Lane limit information (1 of 3 lanes is currently unavailable)

  Further, the accident forecast table creation device 3 in FIG. 1 may be clouded by using cloud computing technology.

  Although the embodiments of the present invention have been described above, the above embodiments are merely examples, and are not intended to limit the scope of the invention. The above embodiment can be implemented in various forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. The above-described embodiments are included in the scope and the gist of the invention, and are also included in the invention described in the claims and its equivalent scope.

1, 1a, 1b Accident prediction system 2, 2a Road traffic control device 3, 3a Accident prediction table creation device 21, 21a Processing unit 22, 22a Storage unit 23 Display unit 24 Input unit 211 Traffic data acquisition processing unit 212 Accident prediction processing Part 213 Reception processing part 214 Display control part 221 Current database 222 Accident prediction table database 31 Processing part 32 Storage part 33 Display part 34 Input part 311 Accident prediction table creation processing part 312 Transmission processing part 321 Past database RS Road sensor part

Claims (6)

For each of a plurality of roads, each of which is divided into a plurality of sections, an accident based on a predetermined learning algorithm using past traffic data and past accident data for each section group into which the section is classified based on road conditions. An accident forecast table creation processing unit that learns the occurrence patterns and creates an accident forecast table that represents the likelihood of occurrence of an accident for each traffic situation,
For each section, a traffic data acquisition processing unit that acquires current traffic data from a sensor that measures traffic conditions,
For each of the plurality of roads, the likelihood of an accident is predicted for each section using the current traffic data of the section and the accident prediction table corresponding to the section group into which the section is classified. An accident forecast processing unit that
An accident forecasting system. As the section group, a group of sections having an uphill, a group of sections having a downhill, a group of sections having a curve, a group of sections having a branch, a group of sections having a merge, a group of sections having a tollgate, group section of the tunnel, a group of the section of the bridge, a group of sections with a number of lanes increases portion includes at least either of the groups of sections with a number of lanes decreases locations, accidents according to claim 1 Forecast system.   The accident prediction table creation processing unit uses the past traffic data and the past accident data acquired at that time at a predetermined timing or when a user inputs an instruction to generate the accident prediction table. The accident forecasting system according to claim 1, which is updated. The accident prediction table creation processing unit uses past traffic data and past accident data for each of a plurality of roads divided into a plurality of sections, for each section group into which the sections are classified based on road conditions. An accident forecast table creation processing step of creating an accident forecast table that represents the likelihood of occurrence of an accident for each traffic situation by learning an accident occurrence pattern based on a predetermined learning algorithm.
A traffic data acquisition processing unit, for each section, a traffic data acquisition processing step of acquiring current traffic data from a sensor that measures traffic conditions,
The accident prediction processing unit, for each of the plurality of roads, for each section, the current traffic data of the section, and using the accident prediction table corresponding to the section group into which the section is classified , An accident forecast processing step that forecasts the likelihood of occurrence,
Accident forecast method including. As the section group, a group of sections having an uphill, a group of sections having a downhill, a group of sections having a curve, a group of sections having a branch, a group of sections having a merge, a group of sections having a tollgate, The accident according to claim 4 , which includes at least one of a group of sections with a tunnel, a group of sections with a bridge, a group of sections with an increased lane number, and a group of sections with an increased lane number. Forecast method. The accident forecast table creation processing unit uses the past traffic data and past accident data acquired at that time at a predetermined timing or when a user inputs an instruction to create the accident forecast table. The accident forecasting method according to claim 4 , further comprising an updating step for updating the accident forecasting table.

Images