JP7350546B2 - Disaster situation estimation device, display system, disaster situation estimation method, disaster estimation model creation method and program - Google Patents

Description

The present invention relates to a disaster situation estimation device, a display system, a disaster situation estimation method, a disaster estimation model creation method, and a program.

When a disaster occurs, a lot of information is transmitted from the scene. In recent years, many people have been using SNS (Social Networking Services), and many reports of disasters are sent by ordinary people through SNS. In response to this situation, Patent Documents 1 to 3 disclose methods of estimating the occurrence of a disaster and its location by using information transmitted on SNS. Information sent on SNS may contain a lot of inaccurate information. In contrast, the technology described in Patent Document 1 performs predetermined processing on information transmitted via SNS to remove noise, thereby making it possible to estimate the area where a disaster has occurred with high accuracy.

JP2016-212751A Japanese Patent Application Publication No. 2016-153971 JP2018-92467A

In addition to information reported by the fire department, police, etc., the disaster response headquarters must also respond to information circulating on the Internet, such as SNS. However, the characteristics of this information differ depending on its source. For example, although the information obtained from the fire department and police is accurate, there are delays in reporting. Information on SNS lacks accuracy for the disaster response headquarters to use as material for decision-making, and it is necessary to verify its authenticity.
Disaster control headquarters must process reported information in a short time and respond appropriately to disasters. Therefore, the task of confirming information reported via SNS becomes a heavy burden. There is a need for a method to accurately grasp the situation of a disaster based on disaster information with different characteristics. Patent Documents 1 to 3 do not disclose such a point.

Therefore, an object of the present invention is to provide a disaster situation estimation device, a display system, a disaster situation estimation method, a disaster estimation model creation method, and a program that can solve the above-mentioned problems.

According to one aspect of the present invention, a disaster situation estimating device includes an estimation model that outputs information indicating the disaster situation when disaster report information reported from a plurality of information sources is acquired, and a predetermined estimation model for an actual disaster. the report information reported from a plurality of the information sources up to the estimation target time; an estimating unit that estimates the actual disaster situation at the estimation target time ; and based on regional characteristics and the type of disaster. a disaster situation simulating unit that simulates a virtual disaster situation indicating the situation of the disaster; and a report information generation pattern that generates a plurality of occurrence patterns of the report information based on the virtual disaster situation and predetermined characteristics of the information source. A generation unit, the plurality of virtual disaster situations simulated by the disaster situation simulation unit, and the plurality of occurrence patterns of the report information generated by the report information generation pattern generation unit for each virtual disaster situation. A learning unit that creates an estimation model.

According to one aspect of the present invention, the plurality of information sources in the disaster situation estimating device include a predetermined organization that takes measures against the disaster and an SNS.

According to one aspect of the present invention, the estimation model takes into account the characteristics of the time required from the occurrence of the disaster to the report, which each of the information sources has, and the characteristics of the accuracy of the report information. Estimate the actual disaster situation.

According to one aspect of the present invention, the learning unit creates the estimation model for each elapsed time, which estimates the situation of the disaster according to the elapsed time since the disaster occurred.

According to one aspect of the present invention, the report information generation pattern generation unit may generate the report information so that when the information source is a large organization, the first generation of the report information is delayed than when the information source is a small organization. Generate information occurrence patterns.

According to one aspect of the present invention, when the information source is an SNS, the report information occurrence pattern generation unit is configured to determine the population density of a predetermined area that is the source of the report information, and the occurrence of the disaster in the area. The generation pattern of the report information is generated according to the distance from the place, the attributes of people existing in the area, and the actual disaster situation.

According to one aspect of the present invention, the disaster situation estimating device learns the relationship between information indicating the disaster situation and an occurrence pattern indicating a change over time in the number of occurrences of the disaster-related report information by information source. In the created estimation model, when report information of the disaster reported from a plurality of the information sources is acquired before the predetermined time has elapsed since the occurrence of the disaster, the predetermined time has elapsed since the occurrence of the disaster. the estimation model that outputs information indicating the situation of the disaster at the time, and the report information reported from the plurality of information sources about the actual disaster by the predetermined estimation target time. An estimating unit that estimates the actual disaster situation.

According to one aspect of the present invention, a display system includes a plurality of display sections and the disaster situation estimation device according to any one of the above, and each of the display sections is configured such that the disaster situation estimation device Display information indicating the estimated disaster situation.

According to one aspect of the present invention, a disaster situation estimation method is a disaster situation estimation method executed by a computer, and includes information indicating a disaster situation and the number of occurrences of report information regarding the disaster by information source over time. This is an estimation model created by learning the relationship with occurrence patterns indicating natural changes, and is an estimation model that calculates disaster report information reported from a plurality of information sources until a predetermined time elapses after the disaster occurs . the estimation model that, when acquired, outputs information indicating the status of the disaster when the predetermined time has elapsed from the occurrence of the disaster; and the estimation model that outputs information indicating the situation of the disaster when the predetermined time has elapsed from the occurrence of the disaster, and reports on the actual disaster from the plurality of information sources by the predetermined estimation target time. estimating the actual disaster situation at the estimation target time based on the report information.

According to one aspect of the present invention, there is provided a method of creating a disaster estimation model executed by a computer, the method of creating a disaster estimation model including a virtual disaster that indicates the situation of the disaster based on regional characteristics and the type of disaster. simulating a plurality of situations; generating a plurality of occurrence patterns of report information reported from the information source based on the virtual disaster situation and predetermined characteristics of each of the plurality of information sources; The method includes the step of creating an estimation model showing a relationship between a virtual disaster situation and the occurrence pattern of the plurality of report information generated for each virtual disaster situation.

According to one aspect of the present invention, the program causes the computer to learn the relationship between information indicating the status of a disaster and an occurrence pattern indicating a change over time in the number of occurrences of reported information regarding the disaster by information source. In the created estimation model, when report information of the disaster reported from a plurality of the information sources is acquired before the predetermined time has elapsed since the occurrence of the disaster, the predetermined time has elapsed since the occurrence of the disaster. the estimation model that outputs information indicating the situation of the disaster at the time , and the report information reported from the plurality of information sources about the actual disaster by the predetermined estimation target time. It functions as a means for estimating the actual disaster situation.

According to one aspect of the present invention, the program includes means for causing a computer to simulate a plurality of virtual disaster situations representing the situation of the disaster based on regional characteristics and the type of disaster; Means for generating a plurality of occurrence patterns of report information to be reported from the information source based on predetermined characteristics each has, a plurality of the virtual disaster situations, and a plurality of the report information generated for each of the virtual disaster situations; It functions as a means to create an estimation model that shows the relationship with the occurrence pattern of

According to the present invention, the situation of a disaster can be estimated based on information reported when a disaster occurs.

FIG. 1 is a functional block diagram showing an example of a disaster situation estimating device according to an embodiment of the present invention. FIG. 3 is a diagram illustrating characteristics of report information in an embodiment of the present invention. FIG. 1 is a first diagram illustrating an outline of a disaster situation estimation method according to an embodiment of the present invention. It is the 2nd diagram explaining the outline of the estimation method of the disaster situation in one embodiment of the present invention. FIG. 2 is a first diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 2 is a second diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 7 is a third diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 4 is a fourth diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 5 is a fifth diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 6 is a sixth diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 7 is a seventh diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 8 is an eighth diagram illustrating a method for creating an estimation model in an embodiment of the present invention. FIG. 2 is a first diagram showing an example of a simulation result of a disaster situation when an earthquake occurs in an embodiment of the present invention. FIG. 2 is a second diagram illustrating an example of a simulation result of a disaster situation when an earthquake occurs in an embodiment of the present invention. FIG. 2 is a first diagram showing an example of a generation result of report information when an earthquake occurs in an embodiment of the present invention. FIG. 7 is a second diagram illustrating an example of a generation result of report information when an earthquake occurs in an embodiment of the present invention. FIG. 7 is a third diagram illustrating an example of a generation result of report information when an earthquake occurs in an embodiment of the present invention. FIG. 2 is a first diagram showing an example of a simulation result of a disaster situation when a flood occurs in an embodiment of the present invention. FIG. 7 is a second diagram illustrating an example of a simulation result of a disaster situation when a flood occurs in an embodiment of the present invention. FIG. 2 is a first diagram illustrating an example of a generation result of report information when a flood disaster occurs according to an embodiment of the present invention. FIG. 7 is a second diagram illustrating an example of a generation result of report information when a flood disaster occurs according to an embodiment of the present invention. It is a flow chart which shows an example of processing of a disaster situation estimating device in one embodiment of the present invention. FIG. 1 is a block diagram showing an example of a display system in an embodiment of the present invention. 1 is a diagram illustrating an example of a hardware configuration of a disaster situation estimating device according to an embodiment of the present invention.

<Embodiment>
Hereinafter, a method for estimating a disaster situation and a method for creating an estimation model according to an embodiment of the present invention will be described with reference to FIGS. 1 to 24.
FIG. 1 is a functional block diagram showing an example of a disaster situation estimating device according to an embodiment of the present invention.
As shown in FIG. 1, the disaster situation estimation device 10 includes a disaster situation estimation section 11, an estimation model creation section 12, an input reception section 13, an output section 14, and a storage section 15.

The disaster situation estimating unit 11 estimates the current situation of an actual disaster based on report information reported from a plurality of information sources regarding the actual disaster and a predetermined estimation model. When the predetermined estimation model acquires disaster report information reported from a plurality of information sources, it outputs information indicating the situation of the disaster.
The disaster situation estimation unit 11 includes a disaster report information acquisition unit 111 that acquires report information on disasters that actually occurred, and a disaster report information acquisition unit 111 that inputs the report information acquired by the disaster report information acquisition unit 111 into an estimation model to calculate information on disasters that actually occurred. and an estimating unit 112 that estimates the situation.

The estimation model creation unit 12 creates the above estimation model.
The estimation model creation unit 12 works with a disaster situation simulation unit 121 that simulates a virtual disaster situation that represents a disaster situation based on the characteristics of the region and the type of disaster, and generates an occurrence pattern of report information for the simulated virtual disaster situation. a plurality of virtual disaster situations simulated by the disaster situation simulation unit 121, and a plurality of report information occurrence patterns generated by the report information generation pattern generation unit 122 for each virtual disaster situation. and a learning unit 123 that creates an estimation model showing the relationship.

The input accepting unit 13 accepts input operations by the user.
The output unit 14 outputs the disaster situation estimated by the disaster situation estimation unit 11 to a display device.
The storage unit 15 stores various information necessary for estimating a disaster situation and creating an estimation model.

FIG. 2 is a diagram illustrating characteristics of report information in an embodiment of the present invention.
When a disaster occurs, a large amount of disaster information (hereinafter referred to as report information) is received from multiple information sources. Multiple sources of information include professional organizations that handle disaster preparedness, such as police and fire departments, as well as the general public. Ordinary people sometimes use SNS to report disasters. As shown in FIG. 2, report information has characteristics of accuracy and reporting delay. For example, information reported immediately tends to be less accurate. Information reported by professional organizations is highly accurate, while information reported by ordinary people on SNS is less accurate. Reports by professional organizations tend to be delayed.
In addition, even highly specialized organizations can take too much time to analyze complex events, making it easy to miss the timing of reporting, and when a large amount of report information is received from one point, it is easy for reporting information to be missing. has been confirmed.
Organizing characteristics by information source, information reported by police and fire departments is highly accurate but tends to be delayed, while information reported via SNS is less accurate and tends to report the same events repeatedly. In this embodiment, instead of estimating the disaster situation by analyzing individual report information and extracting highly accurate information, the present embodiment estimates the disaster situation based on the overall trend of the reported information that has been disseminated. Estimate the situation. In other words, although the trends in the occurrence of reported information reflect the characteristics of reporting delays and information accuracy by information source, these characteristics are also recognized as elements that express the disaster situation and are used to estimate the disaster situation. do. Next, the outline of the disaster situation estimation method according to this embodiment will be explained.

FIG. 3 is a first diagram illustrating an outline of a disaster situation estimation method according to an embodiment of the present invention.
(Creation of estimation model)
The upper part of FIG. 3 shows a method of creating an estimation model by the estimation model creation section 12. A large amount of training data is required to create an estimation model that shows the relationship between the overall trend of reported information and the disaster situation. Therefore, in the present embodiment, the disaster situation simulation unit 121 simulates multiple disaster situations by region and disaster type (earthquake, flood, etc.) based on hazard maps published by local governments, disaster resilience information, etc. simulate At this time, the disaster situation simulation unit 121 simulates the disaster situation every time elapsed from the occurrence of the disaster. For example, the disaster situation simulation unit 121 simulates situations such as building collapse for each of a plurality of earthquakes of different scales. The disaster situation simulation unit 121 may simulate multiple patterns of disaster situations by assuming various building collapse rates even for earthquakes of the same magnitude. The disaster situation simulation unit 121 calculates assumed damages 1 to 3 through simulation. Estimated damages 1 to 3 include the number of building collapses and flooding in each district. Note that the disaster situation simulation unit 121 may simulate three or more possible damages.

Next, the report information generation pattern generation unit 122 generates a plurality of report information generation patterns (report patterns 1 to 3, etc.) for each of the simulated assumed damages 1 to 3, etc. For example, the report information generation pattern generation unit 122 generates a generation pattern of report information by the police and a generation pattern of report information by the fire department, based on predetermined rules for each information source that reflect the characteristics of the report information, regarding the assumed damage 1. and a generation pattern of report information by SNS (report pattern 1). At this time, the report information generation pattern generation unit 122 generates a report information generation pattern for each information source for each elapsed time since the occurrence of the disaster. The report information generation pattern generation unit 122 generates report patterns 2 to 3 of report information for the police, fire department, and SNS when the reporting behavior of the police, fire department, and SNS is changed for the assumed damage 1. The report information generation pattern generation unit 122 generates a plurality of report patterns 1 to 3 through simulation. Note that the report information generation pattern generation unit 122 may generate three or more report information generation patterns for one assumed damage.

Next, the learning unit 123 creates an estimation model that shows the relationship between the assumed damages 1 to 3, etc. generated by the above processing and the corresponding report patterns 1 to 3, etc. For example, when the estimation model obtains the number of reports for each information source in report pattern 1 generated for assumed damage 1, it outputs information indicating the disaster situation corresponding to assumed damage 1, and outputs information indicating the disaster situation corresponding to assumed damage 1, Even if the number of reported information for each information source in 3 is obtained, the converter still outputs information indicating a disaster situation corresponding to assumed damage 1. An example of such a converter is shown later in FIG.

Note that the learning unit 123 may create an estimation model for each elapsed time that estimates the disaster situation according to the elapsed time since the disaster occurred. For example, the learning unit 123 calculates the expected damage for 3 hours after the disaster occurrence based on the expected damages 1 to 3 etc. for 3 hours after the disaster occurrence and the reporting patterns 1 to 3 for the respective expected damage up to 3 hours after the disaster occurrence. Create an estimation model for estimating the disaster situation based on the information reported until later. Similarly, the learning unit 123 creates an estimation model that estimates a disaster situation according to an arbitrary elapsed time, such as 6 hours later, 12 hours later, or 24 hours later.

(Estimation of disaster situation using estimation model)
The lower part of FIG. 3 shows a method for estimating a disaster situation by the disaster situation estimating unit 11. The disaster report information acquisition unit 111 acquires report information from the site. The report information includes, for example, report information reported by the police and fire department, and report information sent via SNS. The estimation unit 112 inputs the number of pieces of report information into the estimation model, and obtains information indicating the disaster situation output from the estimation model. Information indicating the disaster situation includes, in the case of an earthquake, the number of building collapses by region, and in the case of water damage, the number of flood damage cases. The disaster response headquarters can determine the next course of action by referring to the disaster situation determined by the disaster situation estimating unit 11.
As explained in the upper part of FIG. 3, the estimation model is based on multiple virtual disaster situations simulated by the disaster situation simulation unit 121 and reports from multiple information sources generated by the report information occurrence pattern generation unit 122 for each virtual disaster situation. It was created by learning the relationship with information generation patterns. Therefore, once report information is acquired, the estimation model takes into account the characteristics of the information source of the acquired report information (for example, the characteristics of the time required from the occurrence of a disaster to the report, the characteristics of the accuracy of the report information), and calculates the actual Estimate the disaster situation.

FIG. 4 is a second diagram illustrating an outline of a disaster situation estimation method according to an embodiment of the present invention.
FIG. 4 shows an example of the estimation model created by the estimation model creation unit 12. The estimation model is composed of converters AC. Here, a matrix whose elements are the number of reports reported and transmitted by each of the n information sources every hour up to k hours after the occurrence of the disaster is prepared, and is used as observation data. For example, when the information source is the police, fire department, or SNS, observation data as shown in the figure can be obtained based on reported information regarding a certain area. Intermediate data X is calculated by applying a converter A represented by a matrix of k rows and k columns from the left to this observed data. This intermediate data X is multiplied by a converter B represented by a matrix of, for example, 1 row and k columns from the left to calculate intermediate data Y. This intermediate data Y is multiplied from the right by a converter C represented by, for example, a matrix of n rows and 1 column to calculate an estimated solution W. The estimated solution W is a value indicating the current disaster situation in a certain district (for example, the number of collapsed buildings in a certain district).
Note that the estimation model illustrated in FIG. 4 is an example. The learning unit 123 can create an estimation model using any method such as regression analysis, machine learning, or deep learning.

(Details of creation of estimation model)
Next, the estimation model creation process will be explained in more detail.
5 to 12 are first to eighth diagrams respectively illustrating a method for creating an estimation model in an embodiment of the present invention.

(Regional characteristics)
Figure 5 shows map information of the simulation target area. The arrows in the figure indicate rivers, and the area in the lower right corner is the ocean. Other areas are areas where people live, and depending on the area, they may be rural areas or residential areas. City A in the figure is a historically old city. City A includes districts 5, 6, 8, and 9. City B is an emerging city. City B includes districts 1, 2, 3, 4, 7, 14, 15, 16, and 17. City C is a fishing village. City C includes District 10. City D is a seaside industrial district. City D includes districts 11, 12, 13, 18, 19, 20, and 21. In addition, districts 14 to 21 are hilly land and have higher elevations than other districts.

Figure 6 shows the characteristics of districts 1 to 21. Here, Kind indicates the type of urban structure. For example, District 1 (ID=1) can be classified into an urban structure type of Kind=2. The urban structure type is like a distribution type of each type of structure in the area. Figure 7 shows details of urban structure types.

The wooden building in the table of FIG. 7 indicates a wooden building with an old earthquake-resistant design. Wooden construction refers to a wooden building with a new earthquake-resistant design. Non-wooden construction refers to non-wooden buildings. Steel frame indicates a steel frame building. RC indicates a reinforced concrete building. For example, in a district classified as Kind=1, the distribution of structures by type shows that approximately 80% of buildings are old wooden buildings and 20% are non-wooden buildings.

Figure 8 shows the average collapse rate by building structure. The disaster situation simulation unit 121 simulates the number of collapsed buildings in each district based on the collapse rate shown in FIG.

(Characteristics of report information)
Next, the characteristics of report information for each information source will be explained.
1. In the case of an earthquake (A) In the case of fire department Figure 9(a) shows the characteristics of information reported by the fire department in the event of an earthquake disaster. Since the fire department's active force is responsible for rescue and extinguishing fires, it is assumed that their information gathering ability is zero. On the other hand, in the event of an earthquake disaster, much damage information is received through 119 calls, so the fire department's understanding of the local situation is basically based on estimates based on 119 calls. Generally, fire department notifications to the disaster response headquarters are scheduled, so it is assumed that the damage estimated from the 119 call will be reported to the headquarters at the time of the regular report. In addition, initial communication (first report to headquarters) and information processing capacity per hour depend on the scale of each fire department headquarters. Specifically, the larger the scale of the organization (for example, the number of units and groups that make up the fire department headquarters), the higher the information processing ability, but the slower the initial communication. The report information generation pattern generation unit 122 generates a generation pattern of report information by the fire department, based on the following conditions, for example.
(1) 119 calls by residents will be reported immediately after the disaster (total number of collapsed buildings x 10).
(2) The processing capacity per hour is lower for smaller fire departments.
(3) Initial communication is faster for smaller fire departments.
(4) The processing capacity per hour fluctuates more for smaller fire departments.
Figure 9(a) shows an assumed example of information processing capacity per hour and initial transmission delay for fire departments in cities A to D. For example, in city A, the number of damage information that can be processed from residents etc. per hour is 300, which is the number of reports from the time the fire department is first contacted until the first report is notified to the disaster control headquarters. Assume that the delay time is 2 hours.

(B) Police Case Figure 9(b) shows the characteristics of information reported by the police during an earthquake disaster. In general, reports of disaster damage are mainly made by calling 119, and 110 calls are rare. Therefore, in this embodiment, it is assumed that the police will mainly understand the damage through visual confirmation reports from patrol vehicles. In general, since police communications to the disaster control headquarters are scheduled, it is assumed that the damage estimated by patrols at that time will be reported to the headquarters at the time of the regular report. In the case of the police as well, it is assumed that the larger the size of the organization (for example, the number of people in each unit or group that makes up the police organization), the larger the number of patrol vehicles and the delay in initial communication. The report information generation pattern generating section 122 generates a generation pattern of report information by the police based on the following conditions, for example.
(1) The patrol route length is set as SQRT [number of sections (houses, etc.) × area of area], and all vehicles perform the search simultaneously.
(2) The smaller the police force, the lower its search capacity per hour.
(3) Initial communication is faster in smaller police departments.
(4) The smaller the police force, the greater the variation in search capacity per hour.
FIG. 9(b) shows an assumed example of the number of patrol vehicles and initial communication delay for the police in cities A to D. For example, in city A, the police patrol with 10 vehicles, and the reporting delay time from when they first obtain disaster information through patrols to when they notify the disaster control headquarters of the first report information is 2. Assume that it is time.

(C) In the case of SNS Figure 10(a) shows the characteristics of information reported by the police during an earthquake disaster.
Based on past disaster cases, it is thought that the transmission of information via SNS is slower in areas with greater damage. In addition, areas where severe damage has occurred tend to have more related damage, and the amount of communications related to that damage itself tends to increase. The report information generation pattern generation unit 122 generates a generation pattern of report information by SNS based on the following conditions, for example.
(1) The number of target callers is assumed to be (number of collapsed houses x 10).
(2) The volume of calls before peak hours increases by twice every hour.
(3) The amount of calls after peak hours decreases by a factor of 2/3 per hour.
(4) Most of the messages sent via SNS come from areas where the damage has not been severe or from densely populated areas. The number of calls from areas with many elderly people is small. (For example, the number of people targeted for communication assumed in (1) is corrected based on the population density and the proportion of elderly people in cities within a predetermined area around the disaster location. Please note that population density refers to evacuation This value is set taking into consideration the number of people gathered in a place, the degree of crowding, the degree of congestion in downtown areas, etc.For example, the population density of city A is determined by the number of people gathered in a place, the degree of crowding, the degree of congestion in downtown areas, etc. The value obtained by multiplying the number of people visiting the downtown area by a predetermined coefficient, or the value obtained by multiplying the number of people gathered at evacuation centers or downtown areas in city A by a predetermined coefficient, and these values. (You can use the value obtained by dividing the city by the area of the city.)
Fig. 10(a) shows an example of the expected peak time of calls and the number of calls depending on the scale of damage. For example, when large-scale damage occurs, it is assumed that five hours after the disaster occurs, the number of calls per person at that time is 10 calls/hour. FIG. 10(b) shows the number of building collapses in a large-scale case and the number of building collapses in a medium-scale case. If the number of collapses is 50 or more, the report information generation pattern generation unit 122 generates a generation pattern of report information by SNS using the peak time of large-scale damage and the number of calls shown in FIG. 10(a).

2. In the case of flood damage (A), (B) Fire and police departments In the case of flood damage, both the fire department and the police take similar actions. Specifically, rescue operations based on 119 and 110 calls, calls for evacuation using patrol vehicles, and investigations of flooded areas will be carried out. Therefore, in this embodiment, it is assumed that a fire engine and a police car patrol the city area and check for flooded areas. The report information generation pattern generation unit 122 generates a generation pattern of report information by the fire department and the police, based on the following conditions, for example.
(1) If the altitude of the search area is less than the altitude of the flooded area, the dwelling units included in the search area = flood damage.
(2) If the altitude of the search area > the altitude of the flooded area, the flood damage is the flood damage detected by the search.
For example, in the case of the police, the assumption shown in FIG. 9(b) can be used for the number of patrolling vehicles and the initial transmission delay. As for the fire department, assumptions are also made based on the number of vehicles the fire department owns. In addition, the initial communication delay of the fire department may be assumed in the same manner as in FIG. 9(a).

(C) In the case of SNS The number of SNS transmissions shall be proportional to the number of flood damage. The report information generation pattern generation unit 122 calculates the number of calls per person at the peak time by multiplying the number of flood damage based on the above relationship of search area < flooded area by a predetermined constant, for example. Further, regarding the peak time, a predetermined time is set depending on the number of flood damage.

(Contents of assumed disaster)
1. Earthquake patterns Two types of earthquakes are assumed: inland earthquakes and subduction zone earthquakes.
(a) Inland earthquakes The expected collapse rate in coastal areas is 1/5, and in mountainous areas 1/2.
(b) Subduction zone earthquake Damage to inland areas is assumed to be 1/2.

2. Flood Damage Patterns Four types of rainfall area patterns are assumed. Precipitation area patterns 1 to 4 are shown in FIGS. 11 and 12.
The left diagram in FIG. 11 shows precipitation area pattern 1. Precipitation area pattern 1 is an example in which precipitation is concentrated near district 1.
The right diagram in FIG. 11 shows precipitation area pattern 2. Precipitation area pattern 2 is an example in which precipitation is concentrated near districts 3, 4, 7, 15, and 16.
The left diagram in FIG. 12 shows precipitation area pattern 3. Precipitation area pattern 3 is an example in which precipitation is concentrated in districts 3, 4, 7, 8, 9, 14 to 21.
The right diagram in FIG. 12 shows precipitation area pattern 4. Precipitation area pattern 4 is an example in which precipitation is concentrated in areas below the page of districts 8 and 9 and districts 5 to 7.

(Simulation of disaster situation and report information)
Assuming that any of the above-mentioned disasters will occur, an example of a disaster situation simulated by the disaster situation simulation unit 121 and a report pattern generated by the report information occurrence pattern generation unit 122 based on the disaster situation is shown below. Shown below.
1. Simulation of Earthquake Disaster Situation The disaster situation simulation unit 121 generates an urban structure pattern for each district based on the urban structure pattern (Kind) for each district shown in FIG. 6 and the urban structure type shown in FIG. The number of buildings in each district is given. The disaster situation simulation unit 121 generates random numbers centering on the proportions occupied by buildings of each structure shown in the plot characteristics shown in FIG. 7, and sets the proportions of old wooden buildings, new wooden buildings, and non-wooden buildings in district 1 (Kind=2 in the case of). The disaster situation simulation unit 121 simulates the number of buildings by multiplying the predetermined number of buildings in district 1 by a set ratio. The results thus obtained are shown in FIG.
FIG. 13 is a first diagram showing an example of a simulation result of a disaster situation when an earthquake occurs in an embodiment of the present invention. In District 1, the number of old wooden buildings, new wooden buildings, and non-wooden buildings are 534, 131, and 117.

FIG. 14 is a second diagram showing an example of a simulation result of a disaster situation when an earthquake occurs in an embodiment of the present invention.
Next, the disaster situation simulation unit 121 generates random numbers for the collapse rate shown in FIG. 8 to calculate the collapse rate. For example, assume that the collapse rate of old wooden buildings follows a Gaussian distribution centered on 0.05, and then calculate a predetermined collapse rate on the Gaussian distribution by generating random numbers using the Monte Carlo method, etc. to reduce variations in the collapse rate. give. The disaster situation simulation unit 121 multiplies the calculated collapse rate by the 534 old wooden buildings shown in FIG. 13 to calculate the number of collapsed old wooden buildings (16) in district 1. The disaster situation simulation unit 121 similarly calculates the number of collapses for new wooden structures, non-wooden structures, steel structures, and RC structures in District 1. The disaster situation simulation unit 121 also calculates the number of collapsed buildings of each structure for districts 2 to 21. The calculation results are shown in FIG. 14(a). FIG. 14A shows the number of collapses by district and by structure simulated by the disaster situation simulation unit 121 assuming, for example, an inland earthquake. Figure 14(b) shows the results of aggregating the number of collapses calculated for each district for cities A to D.

Next, simulation results of report information generated in response to the above earthquake disaster are shown in FIGS. 15 to 17.
FIGS. 15 to 17 are first to third diagrams showing examples of generation results of report information at the time of an earthquake in an embodiment of the present invention, respectively. In the graphs shown in FIGS. 15 to 17, the vertical axis represents the number of reports, and the horizontal axis represents the time elapsed since the disaster occurred.
FIG. 15 shows a generation pattern of report information by the fire department generated by the report information generation pattern generation unit 122 based on the assumptions described with reference to FIG. 9.
FIG. 16 shows the generation pattern of police report information generated by the report information generation pattern generation unit 122 based on the assumptions described with reference to FIG.
FIG. 17 shows an SNS report information generation pattern generated by the report information generation pattern generation unit 122 based on the assumptions described with reference to FIG. 10. FIG. 17 simulates the number of SNS messages sent in response to earthquake disasters that occurred in each of cities A to D.
For example, the report information generation pattern generation unit 122 generates a report information generation pattern for cities A to D such that the larger the scale of the fire department or police organization is, the later the first report is after the occurrence of a disaster.
Further, for example, the greater the damage caused by the disaster, the more the report information generation pattern generation unit 122 increases the number of SNS transmissions and delays the time when the number of transmissions reaches its peak. In addition, the report information generation pattern generation unit 122 generates report information by SNS according to the population density and age distribution (person attributes) of each city for the areas (cities A to D) that are the source of the SNS. Generate a pattern. For example, the report information generation pattern generation unit 122 increases the number of SNS transmissions in a city with a higher population density. Further, for example, the report information generation pattern generation unit 122 calculates a smaller number of SNS transmissions for cities where the population distribution of the city is large in elderly households. Further, for example, the report information occurrence pattern generation unit 122 generates a report information occurrence pattern according to the distance from the place where the disaster occurred. For example, the number of SNS transmissions from cities within a predetermined distance from the disaster location may be calculated from cities outside the range (for example, cities that are far away, or cities where the disaster occurred if the damage is severe). ) will increase the number of SNS calls from
Note that the report information generation pattern generation unit 122 generates variations in the various conditions (processing capacity, initial transmission delay, number of patrolling vehicles, peak time of transmission, etc.) assumed in FIGS. 9 and 10, assuming that each follows a Gaussian distribution. to generate the occurrence pattern of report information.

When the learning unit 123 acquires the number of pieces of report information illustrated in FIGS. 15 to 17, it creates an estimation model that outputs the number of earthquake damage illustrated in FIG. 14.

2. Simulation of Flood Disaster Situation FIGS. 18 and 19 are a first diagram and a second diagram illustrating an example of a simulation result of a disaster situation when a flood occurs according to an embodiment of the present invention.
For example, the graph in FIG. 18 simulates the temporal change in the inundation rate (area ratio) of each district in the case of rainfall area pattern 1 (left diagram in FIG. 11). A district where the inundation rate = 0 is a district where no inundation is observed. This graph shows that, for example, flooding progresses in districts 1 and 5 from time 0, followed by flooding in districts 6 and 7 located downstream, and then further delayed in districts 8 and 9 located further downstream. This shows how flooding occurs.
The disaster situation simulation unit 121 simulates the flooding situation in each district based on the elevation of each district (not shown) and the predetermined amount of precipitation assumed in the rainfall area pattern 1.

The graph in FIG. 19 is a graph that tabulates the number of flooded buildings in each district over time, calculated according to the assumed flooded area, for cities A to D.

Next, simulated results of report information generated in response to the above-mentioned flood disaster are shown in FIGS. 20 and 21.
20 and 21 are a first diagram and a second diagram illustrating an example of the generation result of report information when a flood disaster occurs according to an embodiment of the present invention.
FIG. 20 shows the generation pattern of report information by fire department and police generated by the report information generation pattern generation unit 122 based on the above assumption.
The report information generation pattern generation unit 122 calculates the number of flooded buildings illustrated in FIG. 19 as flood damage, and generates a report pattern for reporting the flood damage calculated by the fire department and police. For example, regarding the initial transmission delay time, the report information generation pattern generation unit 122 generates random numbers to cause variations in the time illustrated in FIG.
FIG. 21 shows a generation pattern of report information by SNS generated by the report information generation pattern generation unit 122 based on the above assumption.
The report information generation pattern generation unit 122 generates a generation pattern of report information by SNS by multiplying the number of flooded buildings illustrated in FIG. 19 by a predetermined constant to calculate the number of SNS transmissions.

When the learning unit 123 acquires the number of pieces of report information illustrated in FIGS. 20 and 21, it creates an estimation model that outputs the number of flooded buildings due to flood damage illustrated in FIG. 19.

Next, the flow of the estimation model creation process and the damage situation estimation process by the disaster situation estimating device 10 will be explained.
FIG. 22 is a flowchart illustrating an example of processing of the disaster situation estimating device according to an embodiment of the present invention.
The storage unit 15 stores information indicating regional characteristics (city structure type, collapse rate, altitude, etc.) for each type of disaster, as well as assumed disasters (earthquakes and their types, flood damage and rainfall area patterns, etc.). do. In addition, the disaster situation estimation device 10 is installed at the disaster control headquarters, and the disaster control headquarters receives disaster reports from the fire department and police when a disaster occurs, and also monitors SNS for report information sent on SNS. Reports shall be received from organizations, etc.

The user performs an operation to instruct the disaster situation estimation device 10 to create an estimation model. At this time, the user specifies the area where the disaster has occurred. The input receiving unit 13 accepts the input of this operation and instructs the estimation model creation unit 12 to create an estimation model. In the estimation model creation unit 12, the disaster situation simulation unit 121 simulates a disaster situation (step S11). The disaster situation simulation unit 121 simulates disasters and damage occurring in the disaster occurrence area specified by the user for each type of damage assumed. The disaster situation simulation unit 121 may simulate changes in the disaster situation over time, as illustrated in FIGS. 18 and 19. The disaster situation simulation unit 121 records the simulated results in the storage unit 15. It is desirable that the disaster situation simulator 121 simulates a plurality of disaster situations (for example, 100 patterns) by varying the collapse rate and amount of precipitation, for example, even for earthquakes of the same type and floods of the same rainfall area pattern.

Next, the report information occurrence pattern generation unit 122 generates a plurality of report information occurrence patterns for each of a plurality of information sources for each disaster situation simulated by the disaster situation simulation unit 121 (step S12). The occurrence pattern of report information is information indicating a change over time in the number of occurrences of report information for each information source, as illustrated in FIGS. 15 to 17 and the like. For example, the report information occurrence pattern generation unit 122 determines that for the same type of earthquake, if residents receive 10 times as many reports as the number of collapses, when 9 times as many reports are received, or when 11 times as many reports are received as the number of collapses, Generate multiple occurrence patterns by slightly varying the occurrence conditions of report information. In addition, the report information generation pattern generation unit 122 generates a disaster situation that simulates the generation patterns of report information by the police, fire department, and SNS while varying the various conditions described with reference to FIGS. 9 and 10. Generate multiple copies for each.

Next, the learning unit 123 creates an estimation model for each elapsed time (step S13). For example, the learning unit 123 selects the disaster situation three hours later from among all disaster situations simulated for earthquakes and floods occurring in cities A to D, and from among the occurrence patterns of a plurality of report information generated for each disaster situation. Report information up to three hours after the disaster occurrence is extracted, and an estimation model is created that shows the relationship between the two three hours after the disaster occurrence. The learning unit 123 creates an estimation model for each elapsed time from the disaster set by the user (for example, 3 hours, 12 hours, 24 hours, 72 hours, etc.). The learning unit 123 records the created estimation model in the storage unit 15.

Next, when a disaster actually occurs, the input reception unit 13 determines whether an estimation instruction for the actual disaster has been acquired (step S14). For example, when a user inputs the number of reports for each elapsed time by the fire department, police, and SNS when a disaster occurs, and performs an operation to instruct disaster estimation while inputting the estimated elapsed time since the disaster, the input reception unit 13 , it is determined that an instruction for estimating the disaster situation has been obtained. If it is determined that the information has been acquired (step S14; Yes), the input reception unit 13 outputs the number of reports by the fire department, police, and SNS to the disaster situation estimation unit 11. In the disaster situation estimation unit 11, the disaster report information acquisition unit 111 acquires report information and elapsed time from the scene (step S15), and formats the report information into the observation data format shown in FIG. 4, for example. Next, the estimation unit 112 estimates the disaster situation (step S16). For example, the estimating unit 112 receives observation data and elapsed time from the disaster situation estimating unit 11, and inputs the observed data into an estimation model corresponding to the elapsed time. The estimation model outputs a disaster situation that corresponds to the occurrence pattern of actual reported information. Based on the output of the estimation model (for example, the graph in FIG. 19), the estimation unit 112 specifies that the result is a flood in the case of rainfall area pattern 1. For example, the storage unit 15 records disaster situation simulation results by the disaster situation simulation unit 121 for each disaster pattern (earthquake type and precipitation area pattern), and the estimation unit 112 refers to this information and Identify patterns. The output unit 14 displays the disaster situation and precipitation area pattern estimated by the estimation unit 112 on the display device (step S17). Similarly, in the case of an earthquake, the estimation unit 112 identifies the type of earthquake (inland earthquake, subduction zone earthquake). The output unit 14 displays the identified earthquake type and the number of collapses by city or district on the display device. The disaster control headquarters will take measures depending on the scale and type of disaster displayed.

Next, when a predetermined period of time has elapsed, the user inputs the report information collected up to that time into the disaster situation estimating device 10. The disaster situation estimation unit 11 estimates the disaster situation using an estimation model corresponding to the elapsed time up to that point. The disaster situation estimating device 10 repeatedly executes the processes from step S14 to step S17 based on the user's operation while taking measures against the disaster.

According to the present embodiment, regarding report information with different characteristics transmitted from a plurality of information sources when a disaster occurs, attention is focused not on individual report information but as a data group. That is, the respective tendencies (biases) of information reported by the fire department, police, and SNS are considered in advance, and the situation of a disaster in which reports containing such biases and errors are disseminated is estimated. This makes it possible to estimate the situation of a disaster without having to examine each piece of information in a time when prompt countermeasures are required. In addition, the disaster response headquarters is required to grasp the scale of the disaster and the place where it has occurred in general rather than the detailed situation, but according to the disaster situation estimating device 10, the estimation results output by the estimation model can be used in various ways. By comparing the results with simulation results of disaster situations conducted assuming various situations, it is possible to get a comprehensive understanding of disaster patterns and disaster scales. In addition, in addition to SNS information, the estimation model is created by adding information reported by specialized organizations such as fire department and police, so it is possible to create a highly accurate estimation model based on information reported by specialized organizations. can. In addition to highly accurate reporting information from specialized organizations such as the fire department and police, we also utilize SNS information sent from places that the fire department and police cannot reach, allowing us to collect information that may be missing from the fire department and police alone. The disaster situation can be estimated based on the following.

In addition, even for earthquakes of the same scale and type, the disaster situation differs greatly depending on the characteristics of the district, but by simulating disasters in each district and creating an estimation model for each district, it is possible to adjust the disaster situation according to the characteristics of the district. Able to estimate the situation.

Furthermore, when an actual disaster occurs, it is possible to estimate not only the current situation but also the future disaster situation. For example, the user uses a predetermined method (e.g., estimates based on the report information occurrence patterns that are similar among the report information occurrence patterns generated by the report information occurrence pattern generation unit 122) based on the report information actually collected up to that time. ) to estimate future report information (for example, the number of reports for each elapsed time by the fire department, police, and SNS up to 10 hours later). Then, the user inputs the report information up to 10 hours later and the elapsed time (elapsed time up to the present + 10 hours) to the disaster situation estimation device 10. Then, the disaster report information acquisition unit 111 acquires report information (including estimation) and elapsed time up to 10 hours ahead (step S15), and formats the report information into the observation data format shown in FIG. 4, for example. Next, the estimation unit 112 estimates the disaster situation (step S16). For example, the estimating unit 112 receives observation data and elapsed time from the disaster situation estimating unit 11, and inputs the observed data into an estimation model corresponding to the elapsed time. The estimation model outputs the disaster situation after 10 hours corresponding to the occurrence pattern of report information that will occur up to 10 hours later. As described above, according to the present embodiment, by inputting report information reported from multiple information sources up to the estimation target time (10 hours later) into the estimation model, the disaster situation at the estimation target time can be estimated. Therefore, by setting the estimation target time to a point in the future, it can also be used to predict future disaster situations.

(Cooperation with display system)
FIG. 23 is a block diagram illustrating an example of a display system according to an embodiment of the present invention.
The display system 20 is used for the purpose of supporting decision-making for disaster countermeasures at the disaster response headquarters. As shown in FIG. 23, the display system 20 includes a base unit 25 and a slave unit 26, and further includes an I/F unit 21 and an information It includes a management section 22, an input device 23, and an output device 24. The information management section 22 is connected to the disaster situation estimation device 10.

The parent device 25 includes a main display section 25A and a main operation section 25B. The main display section 25A is, for example, a large display that can be viewed by multiple personnel involved in disaster countermeasures. A large display may have a main screen and a sub screen. The main operation section 25B is a touch panel provided on the screen of the main display section 25A, or a mouse or keyboard connected to the main display section 25A.

The slave device 26 is a PC (personal computer), a mobile terminal, or the like that has a sub-display section 26A and a sub-operation section 26B. The sub-display section 26A is a small display that can be viewed by each person involved in disaster countermeasures. The sub-display section 26A may have a main screen and a sub-screen. The sub-operation section 26B is a touch panel provided on the screen of the sub-display section 26A, or a mouse or keyboard connected to the sub-display section 26A. Note that a plurality of slave devices 26 may exist.

The I/F unit 21 is configured as an input/output unit, and inputs and outputs various data. A slave device 26 is connected to the I/F section 21 . Furthermore, an input device 23 and an output device 24 are connected to the I/F section 21 . The input device 23 is a PC or a mobile terminal, and the output device 24 is a printer, a PC, or the like. The disaster situation estimating device 10 is connected to the I/F section 21 . The disaster situation estimating device 10 outputs the estimation result of the disaster situation to the I/F section 21 .

The information management section 22 creates display data to be displayed on the main display section 25A by processing various data input from the outside. Further, the information management section 22 switches the display on the main display section 25A based on the operation on the main operation section 25B.
Further, the information management section 22 causes the slave device 26 to share the display data to be displayed on the main display section 25A of the parent device 25 and display it on the sub display section 26A based on the operation of the sub-operation section 26B.

The information management unit 22 also collects information indicating the disaster situation from the disaster situation estimation device 10, generates display data, and displays this display data on the main display section 25A of the base unit 25 and the sub display section 26A of the slave unit 26. indicate.

By incorporating the disaster situation estimating device 10 into the display system 20, the disaster situation estimated by the disaster situation estimating device 10 can be shared by a plurality of people. When a disaster occurs, information tends to become confused. By using the display system 20, the same disaster information can be displayed on the display screen of each user's terminal device, so they can communicate while referring to the same information and make decisions regarding disaster countermeasures.

FIG. 24 is a diagram illustrating an example of the hardware configuration of a disaster situation estimating device according to an embodiment of the present invention.
The computer 900 includes a CPU 901, a main storage device 902, an auxiliary storage device 903, an input/output interface 904, and a communication interface 905.
The above-described disaster situation estimation device 10 is implemented in a computer 900. Each of the above-mentioned functions is stored in the auxiliary storage device 903 in the form of a program. The CPU 901 reads the program from the auxiliary storage device 903, expands it to the main storage device 902, and executes the above processing according to the program. Further, the CPU 901 reserves a storage area in the main storage device 902 according to the program. Further, the CPU 901 secures a storage area in the auxiliary storage device 903 to store the data being processed according to the program.

A program for realizing all or part of the functions of the disaster situation estimating device 10 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed. Processing may be performed by a functional unit. Note that the "computer system" herein includes hardware such as an OS and peripheral devices. Furthermore, the term "computer system" includes the homepage providing environment (or display environment) if a WWW system is used. Furthermore, the term "computer-readable recording medium" refers to portable media such as CDs, DVDs, and USBs, and storage devices such as hard disks built into computer systems. Further, when this program is distributed to the computer 900 via a communication line, the computer 900 that received the distribution may develop the program in the main storage device 902 and execute the above processing. Further, the above program may be one for realizing a part of the above-mentioned functions, and further may be one that can realize the above-mentioned functions in combination with a program already recorded in the computer system. .
Moreover, the disaster situation estimating device 10 may be configured by a plurality of computers 900. The storage unit 15 may be stored in an external storage device separate from the computer 900. Further, the disaster situation estimation section 11 and the estimation model creation section 12 may be implemented in separate computers 900.

In addition, it is possible to appropriately replace the components in the above-described embodiments with well-known components without departing from the spirit of the present invention. Further, the technical scope of the present invention is not limited to the above-described embodiments, and various changes can be made without departing from the spirit of the present invention.

10... Disaster situation estimation device 11... Disaster situation estimation unit 111... Disaster report information acquisition unit 112... Estimation unit 12... Estimation model creation unit 121... Disaster situation simulation unit 122... - Report information generation pattern generation section 123...Learning section 13...Input reception section 14...Output section 15...Storage section 20...Display system 21...I/F section 22... Information management section 23...Input device 24...Output device 25...Main device 25A...Main display section 25B...Main operation section 26...Slave device 26A...Sub display section 26B. ...Sub-operation unit 900...Computer 901...CPU
902... Main storage device 903... Auxiliary storage device 904... Input/output interface 905... Communication interface

Claims (12)

An estimation model that outputs information indicating the situation of the disaster when report information of a disaster reported from a plurality of information sources is obtained, and an estimation model that outputs information indicating the situation of the disaster when the report information of the disaster reported from a plurality of information sources is obtained, and an estimation model that outputs information indicating the situation of the disaster, and an estimation unit that estimates the actual disaster situation at the estimation target time based on the report information ;
a disaster situation simulation unit that simulates a virtual disaster situation showing the situation of the disaster based on the characteristics of the region and the type of the disaster;
a report information occurrence pattern generation unit that generates a plurality of occurrence patterns of the report information based on the virtual disaster situation and predetermined characteristics of the information source;
creating the estimation model that shows the relationship between the plurality of virtual disaster situations simulated by the disaster situation simulation unit and the plurality of report information occurrence patterns generated by the report information occurrence pattern generation unit for each virtual disaster situation; A learning club and
Disaster situation estimation device equipped with The plurality of information sources include a predetermined organization that takes measures against the disaster and SNS,
The disaster situation estimation device according to claim 1. The estimation model estimates the actual disaster situation by taking into account the characteristics of the time required from the occurrence of the disaster to the report, which each of the information sources have, and the accuracy characteristics of the report information.
The disaster situation estimation device according to any one of claims 1 to 2. The learning unit creates the estimation model for each elapsed time to estimate the situation of the disaster according to the elapsed time since the disaster occurred.
The disaster situation estimating device according to any one of claims 1 to 3 . The report information generation pattern generation unit generates the report information generation pattern so that when the information source is a large organization, the first generation of the report information is delayed than when the information source is a small organization.
The disaster situation estimation device according to any one of claims 1 to 4 . When the information source is an SNS, the report information occurrence pattern generation unit generates information on the population density of a predetermined area from which the report information is transmitted, the distance from the place where the disaster occurred in the area, and the area generating an occurrence pattern of the report information according to the attributes of the people present in the disaster situation and the actual disaster situation;
The disaster situation estimation device according to any one of claims 1 to 5 . This is an estimation model created by learning the relationship between information indicating the disaster situation and occurrence patterns indicating changes over time in the number of reported information regarding the disaster by information source. The estimation model outputs information indicating the status of the disaster when the predetermined time has elapsed since the occurrence of the disaster, when report information of the disaster reported from a plurality of the information sources is acquired before a predetermined time elapses . and the report information reported from the plurality of information sources by the predetermined estimation target time regarding the actual disaster, an estimation unit that estimates the situation of the actual disaster at the estimation target time;
Disaster situation estimation device equipped with comprising a plurality of display units and the disaster situation estimation device according to any one of claims 1 to 7,
A display system in which each of the display units displays information indicating a disaster situation estimated by the disaster situation estimation device. A disaster situation estimation method executed by a computer, the method comprising:
This is an estimation model created by learning the relationship between information indicating the disaster situation and occurrence patterns indicating changes over time in the number of reported information regarding the disaster by information source. The estimation model outputs information indicating the status of the disaster when the predetermined time has elapsed since the occurrence of the disaster, when report information of the disaster reported from a plurality of the information sources is acquired before a predetermined time elapses . and the report information reported from a plurality of information sources regarding the actual disaster by a predetermined estimation target time, estimating the situation of the actual disaster at the estimation target time;
A disaster situation estimation method that has A method for creating a disaster estimation model executed by a computer, the method comprising:
simulating a plurality of virtual disaster situations representing the disaster situation based on regional characteristics and disaster types;
generating a plurality of occurrence patterns of report information reported from the information source based on the virtual disaster situation and predetermined characteristics possessed by each of the plurality of information sources;
creating an estimation model indicating a relationship between the plurality of virtual disaster situations and the plurality of occurrence patterns of the report information generated for each of the virtual disaster situations;
A method for creating a disaster estimation model with computer,
This is an estimation model created by learning the relationship between information indicating the disaster situation and occurrence patterns indicating changes over time in the number of reported information regarding the disaster by information source. The estimation model outputs information indicating the status of the disaster when the predetermined time has elapsed since the occurrence of the disaster, when report information of the disaster reported from a plurality of the information sources is acquired before a predetermined time elapses . and means for estimating the situation of the actual disaster at the estimation target time based on the report information reported from the plurality of information sources by the predetermined estimation target time regarding the actual disaster;
A program to function as computer,
means for simulating a plurality of virtual disaster situations representing the disaster situation based on regional characteristics and disaster types;
means for generating a plurality of occurrence patterns of report information reported from the information source based on the virtual disaster situation and predetermined characteristics possessed by each of the plurality of information sources;
means for creating an estimation model indicating a relationship between the plurality of virtual disaster situations and the plurality of occurrence patterns of the report information generated for each of the virtual disaster situations;
A program to function as

Images