JP2021068070A - Accident cause prediction device, accident cause prediction system, accident cause prediction method, and accident cause prediction program - Google Patents

Abstract

To enhance accuracy in predicting a cause of an accident having possibility of occurring in a place having an infrastructure facility.SOLUTION: An accident cause prediction device 30 includes: a learning model 312 which indicates a machine learning result of relation between a weather state of a place where an accident has occurred and the cause of the accident on the basis of accident history information 411 indicating a history of the place and the cause of the accident and first weather information 412 indicating the weather state of the place in a prescribed period; and a prediction unit 32 which predicts a cause of an accident having possibility of occurring at the place on the basis of second weather information 311 indicating the weather state of the place in a specific period.SELECTED DRAWING: Figure 7

Description

The present invention relates to a technique for predicting an accident cause of an accident that may occur in an infrastructure facility.

The impact of an accident on social infrastructure equipment on people's lives can be very large depending on the nature of the accident. For example, transmission and distribution lines are laid out in various places to supply electric power to each electric power consumer, and are one of the important infrastructure facilities. If an abnormality such as contact with trees or birds and beasts, equipment failure, or lightning strike occurs on this transmission / distribution line, a power outage may occur, and the power outage may have a great impact on people's lives and economic activities. There is. Therefore, in order to respond appropriately to accidents that occur in such infrastructure equipment and to be able to recover quickly, the cause of the accident that occurred is analyzed with high accuracy, and the occurrence of the accident and its effects are analyzed. Expectations for technology for forecasting are increasing.

As a technique related to such a technique, Patent Document 1 discloses a device for determining whether or not the input accident information matches among the accident information that has occurred in the past. If there is no match, this device learns a neural network that matches the pattern matching between the accident information of the past accident and the accident information of this accident, and responds to the accident information of the past accident with the highest degree of similarity. The cause of the accident is estimated to be the cause of this accident.

Further, Patent Document 2 discloses a system that predicts snow damage on a transmission line and issues an alarm based on meteorological data such as temperature, wind speed, and precipitation. This system corrects the meteorological data of the small area to be alerted to the predicted meteorological data of the small area using the predicted meteorological data of the larger area, and based on the predicted meteorological data of the corrected small area. Predict snow damage.

Further, in Patent Document 3, when quantifying the past accident situation data in which the cause of the accident of the transmission line is known, the accident cause of the accident is also quantified with respect to the accident situation data of qualitative expression. The discrimination method is disclosed. In this method, the characteristics of each accident cause are extracted from the quantified accident situation data by using pattern recognition. In this method, the accident situation data whose accident cause is unknown is subsequently quantified, and the quantified accident situation data using pattern recognition is compared with the characteristics of each accident cause extracted in advance, and has the highest degree of similarity. Find the cause of the major accident and determine the cause of the accident with the highest degree of similarity as the cause of the accident.

Further, Patent Document 4 discloses a method of extracting equipment directly related to the cause of the instantaneous voltage decrease detected at the electric place corresponding to the place designated by the user based on the actual data of the instantaneous voltage decrease. .. In this method, based on the extracted equipment position data, map data that identifies an area that may cause an instantaneous voltage drop at least at a location specified by the user is generated and output.

Japanese Unexamined Patent Publication No. 05-250593 Japanese Unexamined Patent Publication No. 08-194970 Japanese Unexamined Patent Publication No. 2000-184593 Japanese Unexamined Patent Publication No. 2005-0274226

"GIS Home Page (Ministry of Land, Infrastructure, Transport and Tourism, Land Policy Bureau, Land Information Division)", [Online], [Search on October 8, 2019], Internet <URL: http://nlftp.mlit.go.jp/ksj/gml/ datalist / KsjTmplt-G02.html> "scikit-learn Machine Learning in Python", [Online], [Searched October 8, 2019], Internet <URL: https://scikit-learn.org/stable/>

Accidents that occur in infrastructure equipment such as transmission and distribution lines occur due to various causes. The cause of the accident is considered to be largely dependent (related) to the characteristics of the place (region) where the infrastructure equipment is located and the weather conditions at that place.

For example, when a large amount of rain falls in a place with a large river or in a mountainous area, accidents such as disruption of railways and roads and collapse of power transmission towers due to floods and landslides may occur. Alternatively, for example, if a strong wind is generated in a place where the transmission / delivery electric wires are drawn in the vicinity of a large number of trees, an accident may occur in which the trees come into contact with the transmission / distribution electric wires.

However, the techniques shown in Patent Documents 1 to 4 are sufficient to analyze and predict the cause of an accident based on the relationship between the characteristics of the place (region) where the infrastructure equipment is located and the weather condition at that place. I can't say I'm doing it. Therefore, it cannot be said that the accuracy of predicting the cause of an accident by the techniques shown in Patent Documents 1 to 4 is high.

A main object of the present invention is to provide an accident cause prediction device or the like that enhances the accuracy of predicting the cause of an accident that may occur in a place where an infrastructure facility is located.

The accident cause prediction device according to one aspect of the present invention describes the relationship between the weather condition and the accident cause at the place where the accident occurred, the accident history information showing the history of the place and the accident cause, and a predetermined period. Based on a learning model that represents the result of machine learning based on the first meteorological information that represents the meteorological condition of the place, and a second meteorological information that represents the meteorological condition of the place in a specific period. Therefore, a predictive means for predicting the cause of an accident that may occur at the above-mentioned place is provided.

From another point of view of achieving the above object, the accident cause prediction method according to one aspect of the present invention uses an information processing device to determine the relationship between the weather condition and the accident cause at the place where the accident occurred. A learning model that represents the result of machine learning based on accident history information that represents the history of the cause of the accident and the first meteorological information that represents the meteorological state of the location in a predetermined period, and a specific period for the location. Based on the second meteorological information representing the meteorological condition in the above location, the cause of the accident that may occur at the location is predicted.

Further, from the further viewpoint of achieving the above object, the accident cause prediction program according to one aspect of the present invention sets the relationship between the weather condition and the accident cause at the place where the accident occurred as the place and the accident cause. A learning model that represents the result of machine learning based on the accident history information that represents the history of the location and the first weather information that represents the weather condition of the location in a predetermined period, and the weather of the location in a specific period. Based on the second meteorological information indicating the state, the computer is made to execute a prediction process for predicting the cause of an accident that may occur at the location.

Further, the present invention can also be realized by a computer-readable, non-volatile recording medium in which the accident cause prediction program (computer program) is stored.

The invention of the present application makes it possible to improve the accuracy of predicting the cause of an accident that may occur in a place where infrastructure equipment is located.

It is a block diagram which shows the structure of the accident cause prediction system 1 which concerns on 1st Embodiment of this invention. It is a figure which illustrates the content of the accident history information 211 which concerns on 1st Embodiment of this invention. It is a figure which illustrates the content of the teacher data 220 which concerns on 1st Embodiment of this invention. FIG. 5 is a flowchart showing an operation in which the machine learning device 20 according to the first embodiment of the present invention generates a learning model 112. It is a figure which illustrates the content of the hazard map 130 which concerns on 1st Embodiment of this invention. FIG. 5 is a flowchart showing an operation of predicting an accident cause by the accident cause prediction device 10 according to the first embodiment of the present invention. It is a block diagram which shows the structure of the accident cause prediction apparatus 30 which concerns on 2nd Embodiment of this invention. It is a block diagram which shows the structure of the information processing apparatus 900 which can execute the accident cause prediction apparatus or the machine learning apparatus which concerns on each embodiment of this invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

<First Embodiment>
FIG. 1 is a block diagram showing a configuration of an accident cause prediction system 1 according to a first embodiment of the present invention. The accident cause prediction system 1 is a system that predicts the accident cause of an accident that is likely to occur in, for example, a power transmission / distribution network (transmission / distribution line) for each location (region).

The accident cause prediction system 1 is roughly classified into an accident cause prediction device 10 and a machine learning device 20. The accident cause prediction device 10 and the machine learning device 20 are communicably connected by a communication network such as the Internet or an intranet. The accident cause prediction device 10 may incorporate the machine learning device 20 instead of communicating with the machine learning device 20 which is an external device.

<Configuration and operation of machine learning device 20>
First, the configuration and operation of the machine learning device 20 will be described.

The machine learning device 20 includes a storage unit 21 and a machine learning unit 22. The storage unit 21 is a storage device such as an HDD (Hard Disk Drive) or an electronic memory, and stores accident history information 211 and learning weather information 212. The storage unit 21 is not provided in the machine learning device 20, but may be provided in an external device communicably connected to the machine learning device 20.

FIG. 2 is a diagram illustrating the contents of the accident history information 211. As illustrated in FIG. 2, the accident history information 211 shows the causes of accidents related to each accident that has occurred so far in the power grid for which the cause of the accident is predicted, the place where the accident occurred, and the date and time when the accident occurred. It is information representing and. The accident history information 211 is, for example, created by an electric power company or the like that manages the power transmission and distribution network, and then stored in, for example, an external server device (not shown in FIG. 1). The machine learning device 20 acquires the accident history information 211 from the server device or the like, and then stores the acquired accident history information 211 in the storage unit 21.

The causes of the accident in the accident history information 211 include, for example, incomplete maintenance, incomplete production, incomplete construction, natural deterioration, overload, wind and rain, ice and snow, thunder, earthquake, flood damage, landslide, avalanche, salt, dust, gas, Workers' negligence, intentional or negligence of the public, tree contact, logging, bird and beast contact, other contact with other objects, other accident spillover, fire, etc. are defined.

The accident occurrence location in the accident history information 211 is represented by, for example, (longitude, latitude), but may be represented by an index different from (longitude, latitude).

According to the accident history information 211 illustrated in FIG. 2, for example, at 11:59 on May 22, 2019, at a place where (longitude, latitude) is (145.2443,44.0934), a tree becomes a transmission / distribution line. A contact (tree contact) accident has occurred.

The learning weather information 212 shown in FIG. 1 is, for example, information represented by data published by the Ministry of Land, Infrastructure, Transport and Tourism shown in Non-Patent Document 1. The machine learning device 20 acquires learning weather information 212 from, for example, a server device of the Ministry of Land, Infrastructure, Transport and Tourism (not shown in FIG. 1), and then stores the acquired learning weather information 212 in the storage unit 21. The learning weather information 212 is an example of the first weather information.

The learning weather information 212 is information representing the state of the weather at each place (region) for each predetermined period (for example, month) for the past several decades. The learning weather information 212 includes, for example, precipitation, maximum temperature, minimum temperature, average temperature, sunshine duration, and total solar radiation as elements indicating the state of the weather. The learning weather information 212 may include elements other than the above-mentioned elements as elements indicating the state of the weather. The learning weather information 212 may represent the average value of the numerical values related to the element in the predetermined period. Further, the information indicating the location included in the learning weather information 212 may be, for example, information that can identify a rectangular area represented by the longitude and latitude of the four vertices.

The machine learning unit 22 in the machine learning device 20 generates teacher data 220 based on the accident history information 211 and the learning weather information 212 stored in the storage unit 21.

FIG. 3 is a diagram illustrating the contents of the teacher data 220 according to the present embodiment. The teacher data 220 is information that associates the cause of an accident that occurred at a certain place with the weather condition during a predetermined period (for example, a month) at which the accident occurred at that place. The machine learning unit 22 extracts the accident history for each predetermined period (for example, month) described above from the accident history information 211. The machine learning unit 22 can generate the teacher data 220 illustrated in FIG. 3 by associating the learning weather information 212 with the extracted accident history for a certain month of a certain year at a certain place. is there. Then, the machine learning unit 22 generates teacher data 220 as illustrated in FIG. 3 for each rectangular region represented by, for example, the longitude and latitude of the four vertices described above.

According to the teacher data 220 illustrated in FIG. 3, when an accident caused by contact with a tree shown in the second line occurs, the amount of precipitation in the month in which the accident occurred at the accident location is 101.975 mm (millimeters). Is. Similarly, according to the teacher data 220 illustrated in FIG. 3, the maximum temperature is 13.225 ° C (however, ° C represents the temperature in degrees Celsius), the minimum temperature is 4.1 ° C, and the average temperature is 8 in relation to the accident. The temperature is 3.3 ° C, the sunshine duration is 164,525 hours, and the total solar radiation is 16.675 kW (kilowatt) / m ² (square meter).

After generating the teacher data 220, the machine learning unit 22 can predict the cause of an accident that is likely to occur at a certain place from the weather condition at that place with high accuracy by using the generated teacher data 220. Do machine learning to do. That is, in the process of machine learning, the machine learning unit 22 determines the strength of the relationship between the cause of the accident at a certain place and the individual elements included in the meteorological conditions (that is, precipitation, temperature, sunshine duration, etc.). Generate rules to represent. The machine learning unit 22 performs machine learning using, for example, a random forest classifier of scikit-learn, which is a machine learning library shown in Non-Patent Document 2. The machine learning unit 22 stores the learning model 112 representing the result of machine learning in the storage unit 11 of the accident cause prediction device 10.

Next, with reference to the flowchart of FIG. 4, the operation (process) of the machine learning device 20 according to the present embodiment to generate the learning model 112 will be described in detail.

The machine learning device 20 obtains the accident history information 211 and the learning weather information 212 from an external device, and stores the obtained accident history information 211 and the learning weather information 212 in the storage unit 21 (step S101). The machine learning unit 22 generates teacher data 220 from the accident history information 211 and the learning weather information 212 by associating the time and place included in the accident history information 211 and the learning weather information 212 (step S102). ..

The machine learning unit 22 performs machine learning using the teacher data 220, stores the learning model 112 as a result in the storage unit 11 of the accident cause prediction device 10 (step S103), and completes the entire process.

<Configuration and operation of accident cause prediction device 10>
Next, the configuration and operation of the accident cause prediction device 10 will be described.

The accident cause prediction device 10 shown in FIG. 1 includes a storage unit 11, a prediction unit 12, a generation unit 13, and a display unit 14. The storage unit 11, the prediction unit 12, and the generation unit 13 are, in order, examples of a storage means, a prediction means, and a generation means.

The storage unit 11 is a storage device such as an HDD (Hard Disk Drive) or an electronic memory, and stores the prediction target weather information 111, the learning model 112 described above, and the map information 113. The storage unit 11 is not provided in the accident cause prediction device 10, but may be provided in an external device communicably connected to the accident cause prediction device 10. The forecast target weather information 111 is information input for predicting the accident cause of an accident that may occur in the future in the power transmission and distribution network, and is information representing the weather condition for each region in a specific period (for example, last month). Is. Like the learning weather information 212 described above, the prediction target weather information 111 is also information represented by data published by the Ministry of Land, Infrastructure, Transport and Tourism shown in Non-Patent Document 1, for example.

The prediction unit 12 predicts the accident cause of an accident that may occur for each place (region) based on the prediction target weather information 111 and the learning model 112. For example, assume that the learning model 112 shows a strong relationship between high rainfall and accidents caused by tree contact for a given location. Then, it is assumed that the forecast target meteorological information 111 indicates that the amount of precipitation in the place last month is large. In this case, the prediction unit 12 predicts that there is a high possibility that an accident due to contact with a tree will occur at that location.

As described above, the prediction unit 12 inputs to the generation unit 13 the result of predicting the cause of the accident that may occur at each location.

The generation unit 13 generates a hazard map 130 representing the prediction result input from the prediction unit 12 by using the map information 113 stored in the storage unit 11. FIG. 5 is a diagram illustrating the contents of the hazard map 130 according to the present embodiment. The hazard map 130 is, for example, an accident most likely to occur in each rectangular region of a predetermined size represented by the longitude and latitude of the four vertices included in the map represented by the map information 113. The cause is represented using a display mode (for example, a pattern or a color) that can identify the cause of the accident.

The generation unit 13 displays the hazard map 130 generated as described above on the display unit 14, which is a display, for example.

Next, with reference to the flowchart of FIG. 6, the operation (process) of the accident cause prediction device 10 according to the present embodiment for predicting the accident cause will be described in detail.

The prediction unit 12 predicts the cause of an accident that may occur in each region based on the forecast target weather information 111 representing the weather condition for a specific period (for example, last month) and the learning model 112 for each region. Then, the prediction result is input to the generation unit 13 (step S201). The generation unit 13 generates the hazard map 130 based on the prediction result input from the prediction unit 12 and the map information 113 (step S202). The generation unit 13 displays the generated hazard map 130 on the display unit 14 (step S203), and the entire process ends.

The accident cause prediction device 10 according to the present embodiment can improve the accuracy of predicting the cause of an accident that may occur in a place where the infrastructure equipment is located. The reason is that the accident cause prediction device 10 may occur at each location using a learning model 112 that represents the result of machine learning about the relationship between the weather condition and the cause of the accident that occurs at each location. This is because it predicts the cause of an accident.

The effects realized by the accident cause prediction device 10 according to the present embodiment will be described in detail below.

Accidents that occur in infrastructure equipment such as transmission and distribution lines occur due to various causes. The cause of the accident is considered to largely depend on the characteristics of the place (region) where the infrastructure equipment is located and the weather conditions at that place. Therefore, in order to take measures against accidents that may occur, it is sufficient to predict the occurrence of an accident and the cause of the accident based on the characteristics of the place where the infrastructure equipment is located and the weather conditions at that place. The challenge is to do it.

To solve such a problem, the accident cause prediction device 10 according to the present embodiment includes a storage unit 11 and a prediction unit 12, and operates as described above with reference to, for example, FIGS. 1 to 6. That is, the storage unit 11 describes the relationship between the weather condition at the place where the accident occurred and the cause of the accident, the accident history information 211 showing the history of the place and the cause of the accident, and the weather at the place during a predetermined period. Based on the learning weather information 212 representing the state of, the learning model 112 representing the result of machine learning is stored. Then, the prediction unit 12 predicts the cause of an accident that may occur at the place based on the prediction target weather information 111 and the learning model 112 in a specific period related to the place.

That is, the accident cause prediction device 10 determines the accuracy of predicting the occurrence of an accident and the cause of the accident by using the learning model 112, which is the result of machine learning the relationship between the weather condition and the cause of the accident at each location. Can be enhanced.

Further, the machine learning device 20 according to the present embodiment keeps the cost required for creating a standard (that is, a learning model 112) for predicting the occurrence of an accident and the cause of the accident low by using machine learning. be able to. As a result, the accident cause prediction device 10 according to the present embodiment predicts the cause of an accident that may occur based on a wide variety of combinations of the characteristics of the place where the infrastructure equipment is located and the weather condition at that place. You can do what you do efficiently.

Further, the accident cause prediction device 10 according to the present embodiment generates a hazard map 130 using a display mode capable of identifying the predicted accident cause for each area of a predetermined size indicating each place (region). As a result, the accident cause prediction device 10 according to the present embodiment can present the prediction result of the accident cause at each location to the user in an easy-to-understand manner.

<Second embodiment>
FIG. 7 is a block diagram showing the configuration of the accident cause prediction device 30 according to the second embodiment of the present invention. Similar to the accident cause prediction device 10 according to the first embodiment, the accident cause prediction device 30 according to the present embodiment also places the accident cause of an accident that is likely to occur in infrastructure equipment such as a power transmission and distribution network. It is a device that predicts for each region).

The accident cause prediction device 30 according to the present embodiment includes a prediction unit 32. The prediction unit 32 is an example of a prediction means.

The learning model 312 describes the relationship between the weather condition and the cause of the accident at the place where the accident occurred, the accident history information 411 showing the history of the place and the cause of the accident, and the weather condition of the place during a predetermined period. The result of machine learning is shown based on the first meteorological information 412. The accident history information 411 and the first weather information 412 are, for example, the same information as the accident history information 211 and the learning weather information 212 according to the first embodiment.

The accident cause prediction device 30 may obtain a learning model 312 generated by an external device such as the machine learning device 20 according to the first embodiment from the external device. Alternatively, the accident cause prediction device 30 provides the learning model 312 with a function of machine learning the relationship between the weather condition and the accident cause based on the accident history information 411 and the first weather information 412. You may generate it yourself. The accident cause prediction device 30 may store the learning model 312 in a storage unit (not shown in FIG. 7), for example.

Then, the prediction unit 32 predicts the accident cause of the accident that may occur at the place based on the second weather information 311 and the learning model 312 in a specific period (for example, last month) regarding the place. The prediction unit 32 may display the predicted result on a display device (not shown in FIG. 7).

The accident cause prediction device 30 according to the present embodiment can improve the accuracy of predicting the cause of an accident that may occur in a place where the infrastructure equipment is located. The reason is that the accident cause prediction device 30 may occur at each location using a learning model 312 that represents the result of machine learning about the relationship between the weather condition and the cause of the accident that occurs at each location. This is because it predicts the cause of an accident.

<Hardware configuration example>
In each of the above-described embodiments, each part of the accident cause prediction device and the machine learning device shown in FIGS. 1 and 7 can be realized by a dedicated HW (HardWare) (electronic circuit). Further, in FIGS. 1 and 7, at least the following configuration can be regarded as a function (processing) unit (software module) of a software program including an instruction executed by a processor.
・ Forecasting units 12 and 32,
・ Generation unit 13,
-Display control function in the display unit 14,
-Memory control functions in the storage units 11 and 21
-Machine learning unit 22.

However, the division of each part shown in these drawings is a configuration for convenience of explanation, and various configurations can be assumed at the time of mounting. An example of the hardware environment in this case will be described with reference to FIG.

FIG. 8 is a diagram illustrating an example of a configuration of an information processing device 900 (computer) capable of executing an accident cause prediction device or a machine learning device according to each embodiment of the present invention. That is, FIG. 8 is a configuration of a computer (information processing device) capable of realizing the accident cause prediction device or machine learning device shown in FIGS. 1 and 7, and can realize each function in the above-described embodiment. Represents a hardware environment.

The information processing apparatus 900 shown in FIG. 8 includes the following as components.
-CPU (Central_Processing_Unit) 901,
-ROM (Read_Only_Memory) 902,
・ RAM (Random_Access_Memory) 903,
-Hard disk (storage device) 904,
-Communication interface 905,
・ Bus 906 (communication line),
A reader / writer 908 that can read and write data stored in a recording medium 907 such as a CD-ROM (Compact_Disc_Read_Only_Memory),
-Input / output interface 909 for monitors, speakers, keyboards, etc.

That is, the information processing device 900 including the above components is a general computer in which these components are connected via the bus 906. The information processing device 900 may include a plurality of CPUs 901 or may include a CPU 901 configured by a multi-core processor.

Then, the present invention described by taking the above-described embodiment as an example supplies the computer program capable of realizing the following functions to the information processing apparatus 900 shown in FIG. The function is the above-described configuration in the block configuration diagram (FIGS. 1 and 7) referred to in the description of the embodiment, or the function of the flowchart (FIGS. 4 and 6). The present invention is then achieved by reading, interpreting, and executing the computer program in the CPU 901 of the hardware. Further, the computer program supplied in the device may be stored in a readable / writable volatile memory (RAM 903) or a non-volatile storage device such as a ROM 902 or a hard disk 904.

Further, in the above case, as a method of supplying the computer program into the hardware, a general procedure can be adopted at present. As the procedure, for example, there are a method of installing in the device via various recording media 907 such as a CD-ROM, a method of downloading from the outside via a communication line such as the Internet, and the like. In such a case, the present invention can be regarded as being composed of a code constituting the computer program or a recording medium 907 in which the code is stored.

The invention of the present application has been described above using the above-described embodiment as a model example. However, the invention of the present application is not limited to the above-described embodiment. That is, the present invention can apply various aspects that can be understood by those skilled in the art within the scope of the present invention.

1 Accident cause prediction system 10 Accident cause prediction device 11 Storage unit 111 Meteorological information to be predicted 112 Learning model 113 Map information 12 Prediction unit 13 Generation unit 130 Hazard map 14 Display unit 20 Machine learning device 21 Storage unit 211 Accident history information 212 For learning Meteorological information 22 Machine learning unit 220 Teacher data 30 Accident cause prediction device 311 Second weather information 312 Learning model 32 Prediction unit 411 Accident history information 412 First weather information 900 Information processing device 901 CPU
902 ROM
903 RAM
904 hard disk (storage device)
905 Communication interface 906 Bus 907 Recording medium 908 Reader / writer 909 Input / output interface

Claims (10)

The relationship between the weather condition and the cause of the accident at the place where the accident occurred, the accident history information showing the history of the place and the cause of the accident, and the first weather showing the weather condition of the place in a predetermined period. Accidents that may occur at the location based on a learning model that represents the results of machine learning based on the information and a second weather information that represents the weather conditions for the location over a particular period of time. An accident cause prediction device equipped with a prediction means for predicting the cause of an accident. Further comprising a generation means for generating a hazard map representing one or more prediction results for each of the above locations by the prediction means.
The accident cause prediction device according to claim 1. The generation means generates the hazard map using a display mode capable of identifying the predicted cause of the accident for each region of a predetermined size indicating the location.
The accident cause prediction device according to claim 2. The generation means generates the hazard map showing the cause of the accident most likely to occur in each of the regions.
The accident cause prediction device according to claim 3. A storage means for storing the learning model is further provided.
The accident cause prediction device according to any one of claims 1 to 4. The first meteorological information includes an average value of numerical values representing the state of the meteorological condition in the predetermined period.
The second meteorological information includes an average value of numerical values representing the state of the meteorological condition in the specific period.
The accident cause prediction device according to any one of claims 1 to 5. The accident cause prediction device according to any one of claims 1 to 6.
A machine learning device that generates the learning model,
Accident cause prediction system with. The machine learning device extracts the history in the predetermined period from the accident history information, and uses the extraction result to machine learn the relationship between the weather condition at the place and the accident cause.
The accident cause prediction system according to claim 7. Depending on the information processing device
The relationship between the weather condition and the cause of the accident at the place where the accident occurred, the accident history information showing the history of the place and the cause of the accident, and the first weather showing the weather condition of the place in a predetermined period. Accidents that may occur at the location based on a learning model that represents the results of machine learning based on the information and a second weather information that represents the weather conditions for the location over a particular period of time. Predict the cause of the accident,
Accident cause prediction method. The relationship between the weather condition and the cause of the accident at the place where the accident occurred, the accident history information showing the history of the place and the cause of the accident, and the first weather showing the weather condition of the place during a predetermined period. An accident that may occur at a location based on a learning model that represents the results of machine learning based on the information and a second weather information that represents the weather conditions for a particular period of time for the location. An accident cause prediction program that causes a computer to perform prediction processing that predicts the cause of an accident.

Images