JP7054097B1 - Risk assessment method, evacuation judgment support method, and risk assessment device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To more accurately evaluate the possibility of human damage caused by a disaster. SOLUTION: This is a risk assessment method for evaluating the risk of showing the possibility of human damage due to a disaster for each of a plurality of sections obtained by dividing a predetermined area, and for each of the plurality of sections in the future. Based on the dynamic area information indicating the state of the section that changes at each time of the time, the vulnerability to the human damage is calculated for each time, and the calculated section of the section is calculated for each of the plurality of sections. Includes assessing the risk at a future time of day based on the vulnerability. [Selection diagram] Fig. 1

Description

The present disclosure relates to a risk assessment method, an evacuation decision support method, and a risk assessment device.

In recent years, landslides and river floods due to heavy rains have been increasing. Therefore, a technique for predicting the occurrence of disasters such as sediment-related disasters in advance is being studied. For example, Patent Document 1 discloses that an infrastructure risk index is calculated by evaluating the possibility of collapsed sediment reaching an area through which a railway passes.

Japanese Unexamined Patent Publication No. 2019-70961

However, it is unclear how much human damage will occur if a disaster occurs only from the prediction result of whether or not a disaster will occur.

The present disclosure has been made in view of the above, and an object of the present invention is to provide a technique capable of more accurately evaluating the possibility of human damage caused by a disaster.

In order to achieve the above object, the risk assessment method according to one form of the present disclosure evaluates the risk indicating the possibility of human damage due to a disaster for each of a plurality of sections obtained by dividing a predetermined area. It is a risk assessment method, and for each of the plurality of sections, the vulnerability to the human damage is calculated for each time based on the dynamic area information indicating the state of the section that changes at each time in the future. , For each of the plurality of compartments, including assessing the risk at each future time based on the calculated vulnerability of the compartment.

According to the above risk assessment method, the vulnerability to human damage is calculated for each time based on the dynamic area information, and for each of the multiple sections, the risk is based on the calculated vulnerability of the section. It will be evaluated at each time in the future. Dynamic area information is information related to changes in the area depending on the time of day, and the information is used to calculate the vulnerability. Therefore, it becomes possible to more accurately evaluate the possibility of human damage caused by a disaster.

The calculation of the vulnerability includes the calculation of the vulnerability based on a model constructed by machine learning so as to output the vulnerability in response to the input of the dynamic area information. Can be .

As described above, by configuring the configuration to calculate the vulnerability based on the model built by machine learning, it is possible to calculate the vulnerability more accurately, so that human damage due to a disaster may occur. Sex can be evaluated more accurately.

For each of the plurality of compartments, further including acquiring the risk of the disaster at each future time, assessing the risk is based on the acquired risk and the calculated vulnerability. , The embodiment can include assessing the risk.

As described above, by configuring the risk evaluation based on the risk level of the disaster, it is possible to more accurately evaluate the possibility of human damage caused by the disaster.

The dynamic area information can be in the form of including information indicating the population at each time in the future for each of the plurality of sections.

By configuring the dynamic area information to include information indicating the population, it is possible to calculate the vulnerability to human damage in consideration of the dynamic changes in the population, so that human damage due to a disaster may occur. It becomes possible to evaluate sex more accurately.

The dynamic area information may include information indicating an event at each future time that affects the ease of evacuation in each of the plurality of sections.

By configuring the dynamic area information to include information related to events that affect the ease of evacuation, it is possible to calculate the vulnerability to human damage in consideration of the ease of evacuation, so people due to disasters It is possible to more accurately evaluate the possibility of damage.

The evacuation judgment support method according to one form of the present disclosure is an evacuation unit that is a part of the predetermined area by using the risk evaluated for each of the plurality of sections by the risk evaluation method described in any of the above. It is an evacuation judgment support method that supports the judgment of the necessity of evacuation in the evacuation unit. It includes determining the stage of necessity and outputting the stage of need for evacuation in the evacuation unit.

According to the above evacuation judgment support method, the stage of the necessity of evacuation is determined for each evacuation unit using the risks evaluated by the above risk assessment method. Therefore, after more accurately assessing the possibility of human damage caused by a disaster, the stage of evacuation necessity for each evacuation unit is determined based on the result, so it is based on a more accurate determination result. It is possible to support evacuation decisions.

The risk assessment device according to one form of the present disclosure is a risk assessment device that evaluates the risk of showing the possibility of human damage due to a disaster for each of a plurality of sections obtained by dividing a predetermined area. For each of the plurality of sections, a vulnerability assessment unit that calculates the vulnerability to human damage for each time based on dynamic area information indicating the state of the section that changes at each time in the future, and the plurality of sections. Each section has a risk assessment unit that evaluates the risk at each future time based on the calculated vulnerability of the section.

According to the above risk assessment device, the vulnerability to human damage is calculated for each time based on the dynamic area information, and for each of the multiple sections, the risk is based on the calculated vulnerability of the section. It will be evaluated at each time in the future. Dynamic area information is information related to changes in the area depending on the time of day, and the information is used to calculate the vulnerability. Therefore, it becomes possible to more accurately evaluate the possibility of human damage caused by a disaster.

The present disclosure provides a technique capable of more accurately assessing the possibility of human damage caused by a disaster.

FIG. 1 is a diagram showing an example of a configuration of a disaster risk management system according to one form. FIG. 2 is a diagram showing an example of a hardware configuration of a computer used in a disaster risk management system. FIG. 3 is a diagram showing an example of the functional configuration of the hazard evaluation unit of the disaster risk management system. FIG. 4 is a diagram schematically showing a part of the evaluation contents in the hazard evaluation unit. FIG. 5 is a diagram showing an example of the functional configuration of the vulnerability evaluation unit of the disaster risk management system. FIG. 6 is a diagram schematically showing a part of the evaluation contents in the vulnerability evaluation unit. 7 (a) and 7 (b) are diagrams for explaining the vulnerability evaluation regarding the evacuation route in the vulnerability evaluation unit. FIG. 8 is a diagram illustrating vulnerability assessment combining vital information. FIG. 9 is a diagram showing an example of the functional configuration of the risk assessment unit of the disaster risk management system. FIG. 10 is a diagram schematically showing a part of the evaluation contents in the risk assessment unit. FIG. 11 is a diagram showing an example of the functional configuration of the evacuation judgment support unit of the disaster risk management system. FIG. 12A is a diagram showing an example of a risk contour diagram, and FIG. 12B is a diagram showing an output example of information related to evacuation judgment support. FIG. 13 is a diagram showing a calculation example of information related to evacuation judgment support in chronological order. FIG. 14 is a diagram showing an example of a processing procedure by the disaster risk management system.

Hereinafter, embodiments for carrying out the present disclosure will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are designated by the same reference numerals, and duplicate description will be omitted.

[Disaster risk management system]
FIG. 1 is a diagram showing a configuration of a disaster risk management system 1 according to one embodiment of the present disclosure. The disaster risk management system 1 has a function of evaluating the risk of human damage caused by a disaster in a target area from various information related to the disaster. "Human damage" means that the inhabitants or those who stay in the target area suffer some damage due to the delay in escape, for example, the death, injury, or missing person occurs. Further, the disaster risk management system 1 has a function of generating information to support the determination of the necessity of evacuation of the inhabitants based on the risk evaluation result. The results of the risk evaluation in the disaster risk management system 1 and the information generated in the disaster risk management system 1 to support the judgment of the necessity of evacuation are, for example, related organizations that make judgments on disasters such as government agencies and local governments. Can be provided for. As described above, the disaster risk management system 1 according to the present embodiment also functions as a risk assessment device or an evacuation judgment support device.

The disaster risk management system 1 includes, for example, a weather information acquisition unit 11, a regional information acquisition unit 12, and a social situation information acquisition unit 13 as a part of a functional module. These functional units acquire various information for assessing the risk to a disaster.

The meteorological information acquisition unit 11 has a function of acquiring meteorological information necessary for evaluating a disaster risk. Meteorological information necessary for evaluating disaster risk includes, for example, warning / warning, radar rainfall, flood warning risk distribution, telemeter water level, sediment disaster warning information, precipitation prediction, tide level information, and the like.

The regional information acquisition unit 12 has a function of acquiring information related to the region necessary for evaluating the disaster risk. Area-related information necessary for assessing disaster risk includes, for example, regional topographical distribution, geological distribution, road location information, shelter location information, and road information between the shelter and the target area ( (Including topographical information around the road, etc.)).

The social situation information acquisition unit 13 has a function of acquiring information related to the social situation necessary for evaluating the disaster risk. The information related to the social situation necessary for evaluating the disaster risk includes, for example, information on the inhabitants of each residence (age, presence / absence of evacuees required, etc.), information on the flow of people for each time zone, and the like. It may also include road congestion information that affects the flow of people in the area.

As an example, the disaster risk management system 1 may acquire the above-mentioned various information from the external server 90 by communicating with the external server 90 via the communication network NW. Further, the information listed above is an example, and information other than the information shown above may be acquired.

Further, the disaster risk management system 1 has a hazard evaluation unit 14, a vulnerability evaluation unit 15, a risk evaluation unit 16, a risk contour diagram creation unit 17, an evacuation judgment support unit 18, and an output unit 19 as part of a functional module. Consists of including.

The hazard evaluation unit 14 has a function of evaluating the risk of how much disaster may occur in the target area mainly based on the weather information and the information related to the area. The “hazard” in the disaster risk management system 1 of the present embodiment is an index indicating the degree of risk of a disaster that may cause human damage in the target area. The hazard evaluation unit 14 evaluates the hazard related to the target area from the data related to the target area by using AI (including machine learning and deep learning), a physical model, a statistical model, and the like. Details will be described later.

The vulnerability evaluation unit 15 has a function of evaluating vulnerability to disasters in the target area mainly based on information on the area and information on the social situation. The "vulnerability" in the disaster risk management system 1 of the present embodiment indicates how robust the target area is to disasters. In the event of a disaster, residents are required to evacuate to shelters, etc. in order to prevent human disasters from occurring in the target area. "Vulnerability" can be an index showing how easy it is for people staying in the target area to evacuate to a shelter. In other words, vulnerability can be an index that considers the amount of exposure to disasters. The vulnerability evaluation unit 15 evaluates vulnerabilities related to the target area from data related to the target area using AI (including machine learning and deep learning), a physical model, a statistical model, and the like. Details will be described later.

The risk assessment unit 16 evaluates the risk of human damage caused by a disaster in the target area based on the evaluation results of the hazard evaluation unit 14 and the vulnerability evaluation unit 15. The “risk” in the disaster risk management system 1 of the present embodiment indicates the possibility of human damage caused by a disaster in the target area. In the event of a disaster, the conditions for human damage to the target area are that a disaster that may cause human damage occurs and that people stay in the area (cannot evacuate to a shelter). It will be. Therefore, the risk assessment unit 16 evaluates the risk in consideration of how much the target area can satisfy these conditions. The risk evaluation unit 16 evaluates the risk of human damage related to the target area from the data related to the target area by using AI (including machine learning and deep learning), a physical model, a statistical model, and the like. Details will be described later.

The risk contour diagram creation unit 17 has a function of creating a contour diagram showing the distribution of risks in each region based on the evaluation result by the risk assessment unit 16. The creation of the contour diagram will be described later.

The evacuation judgment support unit 18 generates information for supporting the judgment of the necessity of evacuation by the administrative agency based on the evaluation result by the risk evaluation unit 16 and the risk contour diagram created by the risk contour diagram creation unit 17. For areas where human damage may occur, the government agency will notify the people staying in the area of the information and give step-by-step instructions regarding the need for evacuation in order to prevent the occurrence of human damage. Need to be done. The information generated by the evacuation judgment support unit 18 to support the judgment of the necessity of evacuation is based on the "risk" evaluated by the risk evaluation unit 16 above, and gives an instruction for each area to be an evacuation unit. This is information to support the identification of the stage related to the necessity of evacuation (evacuation order, evacuation advisory, evacuation preparation, evacuation start for elderly people, etc.). The evacuation judgment support unit 18 uses AI (including machine learning and deep learning), a physical model, a statistical model, etc. to determine whether or not evacuation is necessary for each evacuation unit for the target area from the data related to the target area. Generate information to support. Details will be described later.

The output unit 19 has a function of outputting information obtained by processing by the risk assessment unit 16, the risk contour diagram creation unit 17, the evacuation judgment support unit 18, and the like as needed. The output by the output unit 19 may be, for example, configured to be performed in response to a request from an administrative agency or a local government that may be an output destination, or may be configured to periodically output to a specific output destination.

The disaster risk management system 1 may be always in operation, but may be operated by acquiring the weather information or the like when there is a possibility of causing a disaster. In addition, the disaster risk management system 1 has a predetermined time (for example, 10 hours) at a future time (for example, about several hours to several days from the present) to the extent that human damage due to a disaster can be expected. The above hazards, vulnerabilities, and risks are evaluated and evacuation judgment support information is generated for each time zone by dividing it into minutes to several hours. In the following embodiment, the operation of the disaster risk management system 1 when the weather information that may cause human damage is acquired will be described.

FIG. 2 is a diagram showing an example of the hardware configuration of the computer 110 used in the disaster risk management system 1. For example, the computer 110 has a control circuit 100. In one example, the control circuit 100 has one or more processors 101, a memory 102, a storage 103, a communication port 104, and an input / output port 105. Processor 101 executes operating system and application programs. The storage 103 is composed of a hard disk, a non-volatile semiconductor memory, or a storage medium of a removable medium (for example, a magnetic disk, an optical disk, etc.), and stores an operating system and an application program. The memory 102 temporarily stores the program loaded from the storage 103 or the calculation result by the processor 101. The processor 101 functions as an individual functional module by executing a program in cooperation with the memory 102. The communication port 104 performs data communication with another device via the communication network NW according to a command from the processor 101. The input / output port 105 executes input / output of an electric signal to / from an input / output device (user interface) such as a keyboard, a mouse, a monitor, and a touch panel according to a command from the processor 101.

The disaster risk management system 1 may be composed of a plurality of computers 110. Further, when the disaster risk management system 1 is composed of a plurality of computers 110, the plurality of computers 110 may be distributed and arranged, and these may be connected to each other via a communication network NW.

Next, the hazard evaluation unit 14, the vulnerability evaluation unit 15, the risk evaluation unit 16, the risk contour diagram creation unit 17, and the evacuation judgment support unit 18 of the disaster risk management system 1 will be individually described. Hazard evaluation unit 14, vulnerability evaluation unit 15, risk evaluation unit 16, and evacuation judgment support unit 18 combine AI (including machine learning, deep learning, etc.), physical model, statistical model, etc. to evaluate information. -Perform various processes such as generation. The processes performed in each part are independent and different from each other. Further, for example, just as the evaluation result in the hazard evaluation unit 14 is used in the risk evaluation unit 16, the evaluation result in a specific functional unit may be used in another functional unit. In the following embodiment, a case where each part has a functional part assuming that machine learning is performed will be described. In addition, in each part, machine learning is performed to generate a model, and explanations are given assuming that the model is applied to the target data. However, depending on the type of model used in each part, learning using learning data (teacher data) may not be necessary. Further, the trained model may be acquired from the outside and held. Therefore, the functional parts included in each part are changed according to the type of the model used in each part and the method of using the model.

[Hazard Evaluation Department]
As shown in FIG. 3, the hazard evaluation unit 14 includes a learning data acquisition unit 141, a teacher data holding unit 142, a model creation unit 143, a hazard evaluation model holding unit 144, a target data acquisition unit 145, and a model application unit 146. Consists of including.

The learning data acquisition unit 141 has a function of acquiring learning data used for creating a model.

The teacher data holding unit 142 has a function of holding the learning data acquired by the learning data acquisition unit 141 as teacher data.

The model creation unit 143 has a function of creating a model for evaluating hazards based on learning data.

The hazard evaluation model holding unit 144 has a function of holding the model created by the model creating unit 143.

The target data acquisition unit 145 has a function of acquiring data to be evaluated for hazards.

The model application unit 146 has a function of evaluating the hazard of the target data by applying the model to the target data.

The hazard evaluation unit 14 creates and holds a plurality of models for evaluating hazards. When the model is an AI or a statistical model, training data is acquired for each model to be created, and the model is created using the data as teacher data. When the model is a physical model, learning data (teacher data) is unnecessary, and by acquiring and holding a known model, it can be applied to the target data. The created model is held in the hazard evaluation model holding unit 144. When actually performing hazard evaluation, the analysis result is obtained by acquiring the target data and applying the model to the target data. Each functional unit 141 to 146 of the hazard evaluation unit 14 can perform various calculations depending on the type of AI, physical model, statistical model, etc. used in the hazard evaluation unit 14. The hazard evaluation unit 14 may evaluate the hazard for each of a plurality of divisions of the area to be evaluated and at a predetermined time in the future. Examples of the plurality of sections include, but are not limited to, an area in which the target area is divided into a 50 m mesh shape.

FIG. 4 schematically shows a part of the contents of the evaluation performed by the hazard evaluation unit 14. The information to be evaluated is "○○ disaster-related information" such as "flood flood-related information" and "earth and sand disaster-related information". As an example, in the case of flood inundation-related information, for example, information on precipitation (weather information, rainfall prediction information), information on topography of areas where precipitation occurs, information on places where flooding is likely to occur, and the like can be mentioned. In this way, the information to be evaluated is information related to "○○ disaster" (which may cause XX disaster that may cause human damage in the target area).

By combining AI (including machine learning, deep learning, etc.), physical model, and statistical model with these target information, the hazard evaluation result corresponding to "○○ disaster" will be calculated. Examples of the AI used at this time include Deep Learning, Neural Network, Convolutional NN, Random Forest, Boosted Decision Tree, Genetic Algorithm, and the like. In addition, examples of physical models that can be related to river flooding include a rainfall runoff model, a river channel tracking model, a flood flooding model, a sewerage model, and a model that integrates them. Further, as the statistical model, Support Vector Machine ( SVM ), Bayesian Inference, Logistic Regression and the like can be used. From these AI, physical model, statistical model, etc., the model to be used is selected according to the type of target data and the content of the hazard to be evaluated, and the hazard evaluation is performed. When performing one type of hazard evaluation, a plurality of models may be applied.

[Vulnerability Evaluation Department]
As shown in FIG. 5, the vulnerability evaluation unit 15 includes a learning data acquisition unit 151, a teacher data holding unit 152, a model creation unit 153, a vulnerability evaluation model holding unit 154, a target data acquisition unit 155, and a model application. It is configured to include part 156.

The learning data acquisition unit 151 has a function of acquiring learning data used for creating a model.

The teacher data holding unit 152 has a function of holding the learning data acquired by the learning data acquisition unit 151 as teacher data.

The model creation unit 153 has a function of creating a model for evaluating a vulnerability based on learning data.

The vulnerability evaluation model holding unit 154 has a function of holding the model created by the model creating unit 153.

The target data acquisition unit 155 has a function of acquiring data to be evaluated for vulnerability .

The model application unit 156 has a function of evaluating the vulnerability of the target data by applying the model to the target data.

The vulnerability evaluation unit 15 creates and holds a plurality of models for evaluating vulnerabilities. When the model is an AI or a statistical model, training data is acquired for each model to be created, and the model is created using the data as teacher data. When the model is a physical model, learning data (teacher data) is unnecessary, and by acquiring and holding a known model, it can be applied to the target data. The created model is held in the model holding unit 154 for vulnerability evaluation. When actually assessing vulnerabilities, the analysis results can be obtained by acquiring the target data and applying the model to the target data. The vulnerability evaluation unit 15 may evaluate the vulnerability in each of a plurality of divisions in which the area to be evaluated is divided and at a predetermined time in the future. Examples of the plurality of sections include, but are not limited to, an area in which the target area is divided into a 50 m mesh shape.

FIG. 6 schematically shows a part of the contents of the evaluation performed by the vulnerability evaluation unit 15. The information to be evaluated is "static area information (static area information)", "people flow information", and "dynamic area information (dynamic area information)". As an example, "static area information" is basically area information that does not change depending on the time of day, for example, population distribution by age for each residence, residence information (number of floors), evacuation site information. , Information related to roads (width, slope, speed limit), etc. In addition, "human flow information" is information related to the movement of people by day and hour (population information, etc.). For example, many people stay in the city center during the day and stay in a residential area at night. Information on the movement of people by time of day, such as the increase in the number of people, and information on the movement of people on a daily basis, such as the increase in the number of people moving to tourist spots on weekends, can be mentioned. It can be said that human flow information is a kind of dynamic regional information (dynamic regional information). The "dynamic area information" is, for example, fluid area information, and examples thereof include traffic congestion information and traffic regulation information (information such as traffic closure). Traffic congestion information, traffic regulation information, etc. are information that affects the movement of people staying in the area when evacuating, so it can be said that they can affect the ease of evacuation (ease of evacuation). ..

By combining AI (including machine learning, deep learning, etc.), physical models, and statistical models for these target information, "static regional vulnerabilities" and "dynamic regional vulnerabilities" The evaluation result regarding "" will be calculated.

As the AI, the physical model, and the statistical model used at this time, known ones can be used as in the case of hazard evaluation. From these AI, physical model, statistical model, etc., the model to be used is selected according to the type of target data and the content of the vulnerability to be evaluated, and the vulnerability is evaluated. When evaluating one type of vulnerability, a plurality of models may be applied.

In addition, "static regional vulnerability" refers to a vulnerability with little real-time fluctuation. The vulnerabilities of static areas include, for example, the population distribution by age for each dwelling, the number of floors of the dwelling, the distance to the evacuation center, the required time, etc. For example, there is a vulnerability related to the ease of evacuation, which is derived from each (50 m mesh). In addition, the vulnerability of a static area may change depending on, for example, traffic regulation information. FIGS. 7 (a) and 7 (b) show an example of a change in the time required for evacuation due to traffic restrictions and the like. For example, in FIG. 7A, among the evacuation routes R1, R2, R3 from a certain residence H to the evacuation centers A1, A2, A3, the evacuation route R2 to the evacuation center A2 is the route with the shortest travel time. Suppose it was evaluated. Which of the evacuation routes R1 to R3 has the shortest required time can be evaluated by using, for example, a model such as AI. Here, if traffic regulation (traffic closure) occurs on the evacuation route R2, as shown in FIG. 7 (b), it becomes difficult to move using the evacuation routes R2 and R3 toward the evacuation center A2, and the evacuation time becomes become longer. As a result, the evacuation route R1 toward the evacuation center A1 becomes the evacuation route having the relatively shortest required time. In this way, since the vulnerability of a static area is also information that can change with time, at a predetermined timing (for example, when the traffic regulation information is updated or periodically, etc.). It may be a mode to evaluate again.

"Dynamic regional vulnerability" refers to a vulnerability that is expected to cause some real-time fluctuations. As for the vulnerabilities of dynamic areas, for example, the demographic distribution by age, the time required to each evacuation center, and the traffic regulation of the evacuation route (traffic closure, speed regulation, etc.) are combined, and the areas are divided in advance at predetermined intervals. There is a vulnerability that evaluates the ease of evacuation for each area (for example, 250 m mesh). The age-specific vital distribution may be used, for example, in consideration of the fact that the proportion of elderly people / infants can affect the ease of evacuation of humans.

For example, FIG. 8 shows a route to an evacuation center and vital information for each predetermined section (mesh) on a map. By combining this information, it is possible to evaluate the vulnerability (difficulty of evacuating to a shelter) when a resident evacuates for each section. At this time, by using a model such as AI, it is possible to perform an appropriate evaluation in consideration of past achievements and the like. In addition, when evaluating the vulnerability of a dynamic area, not only static area information but also human flow information or dynamic area information can be used.

The processing performed by the vulnerability evaluation unit 15 can be changed, for example, according to the information that can be acquired from the external server 90 or the like. Further, as a method for evaluating the vulnerability in the vulnerability evaluation unit 15, for example, a method for evaluating each section in several ranks may be used. For example, when evaluating the time required for evacuation to an evacuation center as a vulnerability, for example, the required time is calculated for each section using local information (static or dynamic) using a model. It may be configured to give a rank indicating the level of vulnerability according to the calculation result. In this way, the method of evaluating the vulnerability by the vulnerability evaluation unit 15 can be changed as appropriate. Further, similarly to the hazard evaluation unit 14, when evaluating one type of vulnerability, a plurality of models may be applied for evaluation.

[Risk Assessment Department and Risk Contour Diagram Creation Department]
As shown in FIG. 9, the risk assessment unit 16 includes a learning data acquisition unit 161, a teacher data holding unit 162, a model creation unit 163, a risk evaluation model holding unit 164, a target data acquisition unit 165, and a model application unit 166. Consists of including.

The learning data acquisition unit 161 has a function of acquiring learning data used for creating a model.

The teacher data holding unit 162 has a function of holding the learning data acquired by the learning data acquisition unit 161 as teacher data.

The model creation unit 163 has a function of creating a model for evaluating a risk based on learning data.

The risk assessment model holding unit 164 has a function of holding the model created by the model creating unit 163.

The target data acquisition unit 165 has a function of acquiring data to be evaluated for risk .

The model application unit 166 has a function of comprehensively evaluating the risk for each region by applying the model to the target data.

The risk assessment unit 16 creates and holds a plurality of models for assessing risk. When the model is an AI or a statistical model , training data is acquired for each model to be created, and the model is created using the data as teacher data. When the model is a physical model, learning data (teacher data) is unnecessary, and by acquiring and holding a known model, it can be applied to the target data. The created model is held in the risk assessment model holding unit 164. When actually conducting a risk assessment, the analysis results can be obtained by acquiring the target data and applying the model to the target data. The risk assessment unit 16 evaluates the risk for each of a plurality of divisions of the area to be evaluated and at a predetermined time in the future. Examples of the plurality of sections include, but are not limited to, an area in which the target area is divided into a 50 m mesh shape. The unit of the section may be different from the unit used by the hazard evaluation unit 14 or the vulnerability evaluation unit 15.

Further, the risk contour diagram creation unit 17 has a function of creating a risk contour diagram showing the distribution of risks for each region based on the results evaluated by the risk assessment unit 16.

FIG. 10 schematically shows a part of the contents of the evaluation performed by the risk assessment unit 16. The risk assessment unit 16 has a function of evaluating the disaster risk for each section from the evaluation results of the hazard evaluation unit 14 and the evaluation results of the vulnerability evaluation unit 15 as target data.

As described above, the hazard evaluation unit 14 evaluates the hazard based on the information on the amount of precipitation. In the example shown in FIG. 10, a plurality of institutions (for example, A, B, C) that provide precipitation information are assumed, and hazard evaluation results for precipitation information that can be provided by each institution are illustrated. These are the high risk of sediment-related disasters (earth and sand hazards) or the risk of flood disasters for each section (for example, 50 m mesh) that divides the target area into predetermined units. It is an evaluation of whether it is high (flood hazard). As described above, a plurality of hazard evaluation results obtained by the processing in the hazard evaluation unit 14 can be created for each precipitation information that can be provided and for each type of disaster that can occur.

Further, the vulnerability evaluation unit 15 can also evaluate a plurality of vulnerabilities. For example, information on static and dynamic vulnerabilities can be obtained as regional vulnerabilities to sediment-related disasters. Similarly, as for regional vulnerabilities related to flood damage, information on static vulnerabilities and dynamic vulnerabilities can be obtained.

Here, the risk assessment unit 16 obtains individual risks for each of a plurality of disasters by combining these hazard assessment results and vulnerability assessment results. For example, as a risk related to a sediment-related disaster, for each of the precipitation information provided by the institutions A, B, and C, for example, the risk is quantified and calculated for each section (for example, 50 m mesh) divided by a predetermined unit. Similarly, as a risk related to flood damage, for each of the precipitation information provided by the institutions A, B, and C, for example, the risk is quantified and calculated for each section (for example, 50 m mesh) divided by a predetermined unit.

When calculating each risk, for example, the risk for each section can be calculated by multiplying the hazard derived from the precipitation information provided by each institution with the vulnerability related to the disaster. .. As a result, it is possible to obtain data that evaluates the risk for each of a plurality of events and for each type of disaster.

Further, the risk assessment unit 16 calculates the total risk for each section from the data for evaluating the risk for each precipitation information and each type of disaster provided by each institution. The total risk indicates the risk of a disaster occurring in the area within a predetermined time (for example, within 6 hours) from the present. This total risk can be calculated, for example, in consideration of future weather forecasts and individual risks for each of a plurality of events in each section. When calculating the total risk for each section, for example, it is necessary to determine how much each individual risk for each event needs to be considered (weighting). At this time, for example, the total risk may be obtained by applying a model such as AI. Further, the total risk may be configured by redefining a large section (for example, 250 m mesh) as compared with the section when calculating the individual risk, and calculating in the redefined large section unit.

When calculating the total risk, for example, it is assumed that the following method is used. As the first method, statistically using the maximum value, average value, median value, etc. of the risks in 25 50 m meshes (sections in which individual risks are calculated) that can be included in the 250 m mesh when calculating the total risk. The use of indicators can be mentioned. In addition, as a second method, there is a method of estimating the risk by multivariate analysis using machine learning using the result of the expert's determination of the comprehensive risk estimated from the individual risk as teacher data. Further, as a third method, a method of spatially analyzing 25 50 m meshes (sections in which individual risks are calculated) that can be included in the 250 m mesh when calculating the total risk and creating a frequency distribution can be considered. .. In addition, as a fourth method, an expert system (Expert System) is used, and from the individual risks calculated in 25 50 m meshes (sections in which individual risks are calculated) that can be included in the 250 m mesh when calculating the total risk. A method of calculating the total risk according to a certain rule can be considered. Further, as a fifth method, a method of superimposing individual risks by using weighting with reference to past disaster cases can be considered. The above method is an example, and the method is not limited to these, and these methods may be combined.

As the total risk, an index with a large risk weighting to the risk of a specific disaster may be calculated. In this way, the index calculated as the total risk may be appropriately changed in consideration of the circumstances of the output destination local government or the like.

The above-mentioned total risk is calculated (evaluated) for each of a plurality of divisions of the area to be evaluated and for each predetermined time in the future.

The risk contour diagram shows the risks calculated for each section (here, 250 m mesh) in the risk assessment unit 16 corresponding to the positions of the sections. The risk contour map creation unit 17 illustrates the distribution of risks (comprehensive risk) for each region by displaying the information for identifying the risk for each section calculated by the risk assessment unit 16 on the map. Is. As an example, the risk contour diagram is shown in FIG. 12 (a). In order to create a risk contour diagram, it is sufficient to have map information, information for specifying the location (including size) of the section, and information for specifying the total risk for each section.

[Evacuation Judgment Support Department]
As shown in FIG. 11, the evacuation judgment support unit 18 includes a learning data acquisition unit 181, a teacher data holding unit 182, a model creation unit 183, an evacuation judgment support model holding unit 184, a target data acquisition unit 185, and a model application. It is configured to include part 186.

The learning data acquisition unit 181 has a function of acquiring learning data used for creating a model.

The teacher data holding unit 182 has a function of holding the learning data acquired by the learning data acquisition unit 181 as teacher data.

The model creation unit 183 has a function of creating a model for generating information necessary for evacuation judgment support based on learning data.

The evacuation judgment support model holding unit 184 has a function of holding the model created by the model creating unit 183.

The target data acquisition unit 185 has a function of acquiring data for which information for evacuation judgment support is to be generated.

The model application unit 186 has a function of generating information for evacuation judgment support from the target data by applying the model to the target data.

The evacuation judgment support unit 18 creates and holds a plurality of models for generating information necessary for evacuation judgment support. When the model is an AI or a statistical model , training data is acquired for each model to be created, and the model is created using the data as teacher data. When the model is a physical model, learning data (teacher data) is unnecessary, and by acquiring and holding a known model, it can be applied to the target data. The created model is held in the model holding unit 184 for evacuation judgment support. When actually generating information related to evacuation judgment support, the analysis result can be obtained by acquiring the target data and applying the model to the target data. The evacuation judgment support unit 18 generates information related to the judgment of the necessity of evacuation for each "evacuation unit" included in the target area and at each predetermined time in the future. An evacuation unit is an area of a unit that is subject to step-by-step instructions regarding evacuation. The information related to the determination of the necessity of evacuation is the information for specifying the "evacuation stage" indicating the stage of the necessity of evacuation for each evacuation unit.

The evacuation judgment support unit 18 generates reference information for determining the necessity of evacuation in each evacuation unit based on the total risk calculated by the risk assessment unit 16. "Reference information for determining the necessity of evacuation" is, for example, information for a local government to give an instruction regarding evacuation. When giving instructions regarding evacuation, in many cases, for example, instructions are given in units of school districts and towns, not in units of divisions (mesh) used when assessing the above-mentioned risks. Therefore, the evacuation judgment support unit 18 generates information for determining whether or not evacuation is necessary in the area where evacuation is performed, not in the mesh unit, from the above comprehensive risk information.

As a method for calculating information for determining the necessity of evacuation for each evacuation unit (school district unit, etc.) from the total risk information for each mesh unit (identifying the stage related to the necessity of evacuation), for example, It is conceivable to calculate based on the mode or mode (information of the section with the highest risk) of the total risk for each section (mesh) included in the evacuation unit. When the mode is used, it is possible to generate information for determining the necessity of evacuation in consideration of the risk of many areas included in the area constituting the evacuation unit. In addition, when the highest value is used, it is possible to generate information for determining the necessity of evacuation in consideration of the risk of the area having a particularly high risk among the areas constituting the evacuation unit. In addition, for example, using past cases, we created a model to generate information for determining the necessity of evacuation in consideration of areas where human damage is particularly likely to occur among evacuation units, and for the model. By applying the risk of each section, information for determining the necessity of evacuation (information for specifying the “evacuation stage”) may be generated. Information that identifies the evacuation stage includes evacuation at multiple stages, such as evacuation should be done, evacuation preparation is necessary (evacuation preparation), and evacuation may be necessary (warning). This is information indicating the necessity of. This stage is an example, and the number of stages, classification, etc. are changed as appropriate.

FIG. 12A shows the total risk calculated by the risk assessment unit 16 in the risk contour diagram. On the other hand, FIG. 12B is a diagram showing the result of determining the necessity of evacuation in each area where the evacuation determination is actually made. The unit for evaluation is different between FIG. 12 (a) and FIG. 12 (b). Therefore, the evacuation judgment support unit 18 generates information for a person who gives an instruction regarding evacuation, such as a local government, to judge whether evacuation is necessary or not, on a regional basis for determining whether evacuation is necessary (evacuation is performed).

The necessity of evacuation may change from moment to moment due to, for example, changes in the weather (increased precipitation, etc.). Therefore, the information for determining the necessity of evacuation may be generated, for example, every predetermined time zone (for example, every 5 minutes to several hours). In addition, the hazard evaluation results, vulnerability evaluation results, risk evaluation results, etc. may change not only due to changes in the weather but also due to changes in surrounding traffic conditions, etc., so the evaluation results regarding the necessity of evacuation may also change. .. Therefore, for example, the evacuation judgment support unit 18 repeatedly generates information for determining whether or not evacuation is necessary in 1-hour units from the current time to several hours ahead at predetermined time zones (for example, every 10 minutes). (Recalculate) is conceivable. It may be determined in advance at what timing (interval) and in what range (up to how many hours ahead) information for determining whether or not evacuation is necessary is generated.

In addition, in order for the evacuation judgment support unit 18 to generate information for supporting the judgment of the necessity of evacuation, the information related to the comprehensive risk calculated by the risk assessment unit 16 as described above is required. Further, in order to calculate the total risk, the hazard evaluation result in the hazard evaluation unit 14 and the vulnerability evaluation result in the vulnerability evaluation unit 15 are required. Therefore, in order for the evacuation judgment support unit 18 to generate (recalculate) information for supporting the determination of the necessity of evacuation, each of the above units also repeatedly generates necessary information.

FIG. 13 is a diagram showing a state in which the information (estimation result) for determining the necessity of evacuation has changed according to the change in the external information. In FIG. 13, the target area is divided into four evacuation order unit areas, and the result of estimating whether or not evacuation is necessary at each time for each area (generating information for determining whether evacuation is necessary). The result) is shown. In the example shown in FIG. 13, the estimation result of whether or not evacuation is necessary is performed in four stages (“evacuation”, “preparation”, “caution / safety”, and “unnecessary”). FIG. 13 shows the areas where estimation results other than “unnecessary” were obtained among the four stages. Further, FIG. 13 shows a state in which the prediction results in each time zone (10:00 to 18:00) are applied to the map.

Among the prediction results shown in FIG. 13, the result E1 (10:00 to 16:00) shown at the top is an estimation result of the future situation based on the meteorological information and the like currently obtained (at 10:00). On the other hand, the result E2 in the second row from the top is an estimation result after 1 hour has passed (at 11:00). Since the result E2 is an estimation result when the situation is basically unchanged compared to the stage where the result E1 is estimated, the estimation result of the future time zone is almost the same as the result E1 . ..

On the other hand, the result E3 in the third row from the top is an estimated result after another hour has passed (at 12 o'clock). The result E3 shows the result of re-estimating whether or not evacuation is necessary based on the newly obtained information that the traffic was closed due to flooding compared to the stage where the result E2 was estimated. ing. In the estimation result E31 at 12 o'clock of the result E3, the estimation result of the area related to the road closure changes ("unnecessary" → "caution / safety" compared with the estimation result at 12 o'clock in the results E1 and E2. ")are doing. Further, the estimation results after that are different from the results E1 and E2. It is considered that this estimation result is derived from the change in the evaluation result related to the vulnerability related to the evacuation route, for example, due to the occurrence of traffic closure. In addition, it is probable that due to changes in the evaluation results related to vulnerability, the number of areas evaluated as having a higher overall risk increased, and as a result, the estimation result was changed to the presumed result that evacuation is highly necessary.

Further, the result E4 in the fourth row from the top is an estimation result at the same timing (at 11 o'clock) as the result E2. The result E4 shows the result of re-estimating whether or not evacuation is necessary based on the new information that the landslide occurred in the mountainous area as compared with the result E2. .. In the result E4, the estimation result of the area where the landslide occurred is changed (“unnecessary” → “caution / safety”, etc.) as compared with the results E1 and E2. Further, the estimation results after that are also different from the results E1 and E2. This estimation result is considered to be derived from the change in the hazard evaluation result related to the landslide disaster, for example, due to the occurrence of a landslide. In addition, if the evacuation route to a specific evacuation site is hindered as a result of a landslide, it is considered that the evaluation result of the vulnerability related to the evacuation route has changed. Furthermore, it is probable that due to the change in the above evaluation results, the number of areas evaluated as having a high overall risk increased, and as a result, the estimation result was changed to the presumed result that the need for evacuation was high.

As shown in Results E3 and E4, the hazard evaluation result or the vulnerability evaluation result may change depending on the information obtained from the outside. As a result, the overall risk assessment results are changed. As a result, the information (estimated result) for determining the necessity of evacuation may be changed.

[Risk management method in disaster risk management system]
The risk management method by the above-mentioned disaster risk management system 1 will be described with reference to FIG.

In step S01, the weather information acquisition unit 11, the regional information acquisition unit 12, and the social situation information acquisition unit 13 of the disaster risk management system 1 acquire external information from the external server 90, respectively.

In step S02, the hazard evaluation unit 14 evaluates the hazard based on the information acquired by the weather information acquisition unit 11 and the regional information acquisition unit 12. At this time, if necessary, the evaluation is performed using the model held by the hazard evaluation model holding unit 144.

In step S03, the vulnerability evaluation unit 15 evaluates the vulnerability in the area based on the information acquired by the regional information acquisition unit 12 and the social situation information acquisition unit 13. At this time, if necessary, the evaluation is performed using the model held by the vulnerability evaluation model holding unit 154.

In step S04, the risk assessment unit 16 evaluates the risk using the evaluation results of the hazard evaluation unit 14 and the vulnerability evaluation unit 15. At this time, for example, as shown in FIG. 10, a plurality of individual risks are calculated from the evaluation results of the hazard evaluation unit 14 and the vulnerability evaluation unit 15, and further, these are combined to calculate the total risk.

In step S05, the risk contour diagram creation unit 17 creates a risk contour diagram based on the comprehensive risk calculated by the risk assessment unit 16.

In step S06, the evacuation judgment support unit 18 uses the evaluation results of the risk evaluation unit 16 to generate information for supporting the evacuation judgment.

In step S07, the output unit 19 outputs necessary information from the evaluation results. The information to be output can be changed, for example, according to the request of a user (for example, a local government) who uses the system. As an example, when generating information for supporting evacuation judgment, a configuration may be configured in which the evaluation result of the total risk of each mesh unit, which is the basis for generating the information, is combined and output for each evacuation unit. It goes without saying that such an output example is an example and may be changed depending on the output destination.

It should be noted that it is not always necessary to perform the above steps S01 to S07 in this order, and the configuration may be appropriately replaced.

[Action]
As described above, in the risk assessment device (disaster risk management system 1) and the risk assessment method according to the present embodiment, the vulnerability to human damage is calculated for each time based on the dynamic area information, and a plurality of divisions are used. For each of the above, the risk is assessed at each future time based on the calculated parcel vulnerability. Dynamic area information is information related to changes in the area depending on the time of day, and the information is used to calculate the vulnerability. Therefore, it becomes possible to more accurately evaluate the possibility of human damage caused by a disaster.

Further, when calculating the vulnerability, the vulnerability may be calculated based on a model constructed by machine learning so as to output the vulnerability in response to the input of dynamic area information. In this case, since it is possible to calculate the vulnerability more accurately, it is possible to more accurately evaluate the possibility of human damage caused by a disaster.

Further, it may further include acquiring the disaster risk level for each of the plurality of sections at each future time. And if risk assessment includes risk assessment based on acquired risk and calculated vulnerability, more accurately assess the likelihood of human injury from a disaster. be able to.

The dynamic area information may include information indicating the population at each time in the future for each of the plurality of sections. With such a configuration, it is possible to calculate the vulnerability to human damage in consideration of dynamic changes in the population, so it is necessary to more accurately evaluate the possibility of human damage caused by a disaster. Is possible.

The dynamic area information may include information indicating events at each future time that affect the ease of evacuation in each of the plurality of compartments. With such a configuration, it is possible to calculate the vulnerability to human damage in consideration of the ease of evacuation, so it is possible to more accurately evaluate the possibility of human damage caused by a disaster. Will be.

In addition, the evacuation judgment support method according to this embodiment requires evacuation in evacuation units that are part of the area, using the risks evaluated for each of the plurality of sections by the risk evaluation method described in any of the above. It is an evacuation judgment support method that supports sex judgment, and determines the stage of evacuation necessity in the evacuation unit based on the risk of each of the parts corresponding to the evacuation unit among multiple sections. And to output the stage of the need for evacuation in the evacuation unit. With such a configuration, the stage of the necessity of evacuation is determined for each evacuation unit using the risks evaluated by the above risk assessment method. Therefore, after more accurately assessing the possibility of human damage caused by a disaster, the stage of evacuation necessity for each evacuation unit is determined based on the result, so it is based on a more accurate determination result. It is possible to support evacuation decisions.

[Modification example]
As described above, the present disclosure is not necessarily limited to the above-described embodiment, and various changes can be made without departing from the gist thereof.

For example, in the above embodiment, the case where the risk evaluation result by the hazard evaluation unit 14 and the vulnerability evaluation result by the vulnerability evaluation unit 15 are combined to evaluate the risk by the risk evaluation unit 16 has been described. However, this configuration has been described. It can be changed as appropriate. For example, the risk may be evaluated using the result calculated by only one of the hazard evaluation unit 14 and the vulnerability evaluation unit 15. Further, instead of providing the hazard evaluation unit 14 and the vulnerability evaluation unit 15, the configuration may be such that the hazard evaluation result or the vulnerability evaluation result calculated by another device is acquired.

[Additional Notes]
The evacuation judgment support method according to one form of the present disclosure is to evaluate the risk of showing the possibility of human damage due to a disaster for each of a plurality of sections obtained by dividing a predetermined area at each time in the future. And, for the evacuation unit provided in the predetermined area, which includes a part of the plurality of sections, the evacuation for each evacuation unit is based on the risk of each of the part of the sections. It includes evaluating the evacuation stage indicating the stage of necessity for each future time, and outputting the evaluated evacuation stage.

According to the above evacuation judgment support method, it is necessary to evacuate from an evacuation unit based on the risk evaluated for each of the evacuation units, which is composed of a part of a plurality of sections. The evacuation stage indicating the stage of is evaluated and output for each future time. Therefore, it is possible to more accurately evaluate the degree of evacuation necessity.

Evaluating the evacuation stage is based on a model constructed by machine learning to output the evacuation stage in response to the input of the risk of each of the partial sections constituting the evacuation unit. It can be an embodiment that includes evaluating the stage.

As described above, by configuring the evacuation stage to be evaluated based on the model constructed by machine learning, it is possible to evaluate the evacuation stage more accurately, so that the degree of evacuation necessity is more accurate. It becomes possible to evaluate.

Evaluating the evacuation stage includes evaluating the evacuation stage in the evacuation unit based on the mode or mode included in the risk of each of the sections constituting the evacuation unit. It can be an embodiment.

As described above, the evacuation unit is configured by evaluating the evacuation stage in the evacuation unit based on the mode or the highest value included in the risk of each of the partial sections constituting the evacuation unit. Appropriate evaluation can be performed in consideration of the risks of each of the parts to be evacuated.

Evaluating the risk includes assessing the risk for each section based on the risk of the disaster at each future time and the vulnerability to the human injury at each future time. Can be .

As described above, by configuring the risk evaluation based on the risk of disaster and the vulnerability to human damage, it is possible to evaluate the risk more accurately for each section, so evacuation is required in the evacuation unit. Appropriate evaluation can be made regarding the degree of denial.

The evacuation judgment support device according to one form of the present disclosure is a risk of evaluating the risk of showing the possibility of human damage due to a disaster for each of a plurality of sections obtained by dividing a predetermined area at each future time. With respect to the evacuation unit including the evaluation unit and a part of the plurality of sections provided in the predetermined area, from the evacuation unit based on the risk of each of the part of the sections. It includes an evacuation judgment support unit that evaluates the evacuation stage indicating the stage of necessity of evacuation for each future time, and an output unit that outputs the evaluated evacuation stage.

According to the above evacuation judgment support device, for an evacuation unit composed of a part of a plurality of sections, the necessity of evacuation from the evacuation unit based on the risk evaluated for each of the parts. The evacuation stage indicating the stage of is evaluated and output for each future time. Therefore, it is possible to more accurately evaluate the degree of evacuation necessity.

1 ... Disaster risk management system, 11 ... Meteorological information acquisition department, 12 ... Regional information acquisition department, 13 ... Social situation information acquisition department, 14 ... Hazard evaluation department, 15 ... Vulnerability evaluation department, 16 ... Risk evaluation department, 17 ... Risk contour diagram creation department, 18 ... Evacuation judgment support department, 19 ... Output department, 90 ... External server.

Claims (7)

It is a risk assessment method that evaluates the risk of human damage caused by a disaster for each of multiple plots obtained by dividing a predetermined area.
For each of the plurality of sections, the vulnerability to human damage is calculated for each time based on the dynamic area information indicating the state of the section that changes at each time in the future.
For each of the plurality of parcels, the risk should be evaluated at each future time based on the calculated vulnerability of the parcel.
Risk assessment methods, including. The calculation of the vulnerability includes calculating the vulnerability based on a model constructed by machine learning so as to output the vulnerability in response to the input of the dynamic area information. The risk assessment method described in 1. For each of the plurality of sections, the risk of the disaster is further included to be acquired for each future time.
The risk assessment method according to claim 1 or 2, wherein assessing the risk comprises assessing the risk based on the acquired risk and the calculated vulnerability. The risk assessment method according to any one of claims 1 to 3, wherein the dynamic area information includes information indicating the population at each future time for each of the plurality of sections. The risk assessment method according to any one of claims 1 to 4, wherein the dynamic area information includes information indicating an event at each future time that affects the ease of evacuation in each of the plurality of sections. The need for evacuation in evacuation units that are part of the predetermined area using the risks evaluated for each of the plurality of sections by the risk assessment method according to any one of claims 1 to 5. It is an evacuation judgment support method that supports judgment,
To determine the stage of evacuation necessity in the evacuation unit based on the risk of each of the partial sections corresponding to the evacuation unit among the plurality of sections.
Outputting the stage of evacuation necessity in the evacuation unit and
Evacuation decision support methods, including. It is a risk assessment device that evaluates the risk of human damage caused by a disaster for each of a plurality of sections obtained by dividing a predetermined area.
For each of the plurality of sections, a vulnerability evaluation unit that calculates the vulnerability to human damage for each time based on dynamic area information indicating the state of the section that changes at each time in the future.
For each of the plurality of sections, a risk assessment unit that evaluates the risk at each future time based on the calculated vulnerability of the section.
Has a risk assessment device.

Images