JP7382859B2 - Disaster scale estimation device - Google Patents

Description

The present disclosure relates to a disaster scale estimating device that estimates a disaster scale after a disaster occurs.

In recent years, there has been a tendency for disasters caused by wind and rain, such as typhoons and torrential rains, to become more severe. Therefore, there is an increasing need to quickly and accurately estimate the scale of the disaster and the evacuation status of disaster victims after a disaster occurs, and to promptly take appropriate measures such as recovery and support based on the estimation results. Regarding estimating the evacuation situation of disaster victims among the above, a technique for estimating the evacuation situation of disaster victims is proposed in Patent Document 1 below.

Japanese Patent Application Publication No. 2011-086070

On the other hand, when it comes to estimating the scale of a disaster, for example, there are methods to investigate the scale of the disaster using satellite photographs, and methods to estimate the scale of the disaster based on information posted on social networking services (SNS) such as Twitter (registered trademark). There are many possible methods. However, the former method using satellite photographs has problems with the real-time nature of the information, and the latter method based on posted information has problems with the reliability of the information. Therefore, there is a strong need for a method for quickly and accurately estimating the scale of a disaster after it occurs.

Therefore, the present disclosure aims to quickly and accurately estimate the disaster scale after a disaster occurs.

The disaster scale estimating device according to the present disclosure acquires the population data in normal times, which is population data that includes information on the estimated population for each time segment in a predetermined area unit, and uses the acquired population data to A normal population aggregation unit that aggregates the normal population for each time segment in area units, and a normal population aggregation unit that obtains the population data after the occurrence of a disaster, and calculates the population at the time of disaster for each time segment in the area unit from the obtained population data. and a disaster scale estimating unit that estimates the disaster scale based on a predetermined criterion based on the change in the disaster population from the normal population in the same time period in the same area unit.

In the above-mentioned disaster scale estimating device, the normal-time population aggregation unit acquires population data in normal times, which is population data that includes information on the estimated population for each time segment in a predetermined area unit, and Calculate the normal population for each area and time segment from the current population data. Then, the disaster scale estimating unit obtains the population data after the disaster occurs, aggregates the population at the time of the disaster for each time period in the area unit from the obtained population data after the disaster occurs, and calculates the population at the time of the disaster for each time period in the same area unit. Based on the change in the population at the time of the disaster from the normal population, the scale of the disaster is estimated based on predetermined criteria. Regarding the above population data, by making effective use of the fact that accurate population data can be obtained quickly, it is possible to predict the disaster situation after a disaster occurs based on the change in the population at the time of a disaster from the normal population in the same time period in the same area unit. The scale can be estimated quickly and accurately.

According to the present disclosure, it is possible to quickly and accurately estimate the scale of a disaster after it occurs. Accordingly, the results of estimating the scale of the disaster can be effectively utilized, for example, in disaster response by local governments.

FIG. 1 is a functional block diagram showing the configuration of a disaster scale estimating device according to a first embodiment. It is a figure showing an example of population data inputted from the outside. It is a figure showing an example of a normal population table. It is a figure showing an example of a population ratio table. FIG. 3 is a flow diagram showing processing in the first embodiment. FIG. 3 is a flow diagram showing normal population aggregation processing. FIG. 2 is a flowchart showing a population ratio calculation process. FIG. 3 is a flow diagram showing learning processing related to estimation of disaster scale. 9 is a diagram for explaining the learning process of FIG. 8. FIG. FIG. 3 is a flow diagram showing a disaster scale estimation process. 11 is a diagram for explaining the estimation process in FIG. 10. FIG. (a) is a flow diagram showing processing of modification 1 of the first embodiment, and (b) is a flow diagram showing processing of modification 2 of the first embodiment. FIG. 2 is a functional block diagram showing the configuration of a disaster scale estimating device according to a second embodiment. FIG. 7 is a flow diagram showing processing in a second embodiment. FIG. 3 is a flow diagram showing an abnormality score calculation process. FIG. 3 is a flow diagram showing a disaster scale estimation process. It is a diagram showing an example of the hardware configuration of a disaster scale estimating device.

Below, we will discuss the first embodiment in which the disaster scale is estimated based on the "ratio" of the disaster population to the normal population as a change in the disaster population from the normal population, and the first embodiment in which the disaster scale is estimated based on the "ratio" of the disaster population to the normal population. A second embodiment in which the disaster scale is estimated based on the "difference" between the population at the time of disaster and the population during normal times will be described in order.

[First embodiment]
As shown in FIG. 1, the disaster scale estimation device 10 according to the first embodiment includes a normal population aggregation unit 11, a normal population table 12, and a disaster scale estimation unit 13.

The normal population aggregation unit 11 acquires population data in normal times, which will be described later, including information on estimated population for each time segment in a predetermined area unit, from an external server (not shown). This is a functional unit that aggregates the normal population for each area and time segment from the population data obtained. Note that "time" and "weekday/holiday distinction" are collectively referred to as "time division." As the above population data, for example, population data obtained by the population calculation method based on the location information of the user terminal and the user's address attributes, etc. disclosed in International Publication No. WO2012/056900 can be adopted, and as shown in Figure 2. As illustrated, it includes information on the estimated population for each calculation date and time for each area (city, ward, town, village) classified by residential region (for example, A city, B city, C city) based on the user's address attribute. Figure 2 shows only an example of “2020/01/24 09:00:00” as the calculation date and time, but the population data can be calculated using other calculation dates and times, such as “2020/01/24 10:00:00”. Also includes information on estimated population. The normal population counting unit 11 calculates "residents" whose residential area and area are the same, and "residents" whose residential area and area are the same, based on the residential area information and area (in this case, city, ward, town, village) information included in the population data. For each of the different "non-residents" and the "total number" including residents and non-residents, the population average value "average estimated population" and "estimated The dispersion of population is totaled and the total result is stored in the normal population table 12. At this time, as mentioned above, since the "time division" includes not only the time but also the "weekday/holiday distinction," the normal population counting unit 11 may calculate the "holiday (for example, Saturday, holiday)" even if the time is the same. The above estimation is performed by distinguishing between "Sundays and holidays)" and "weekdays (days other than holidays)."

The normal population table 12 is a table that stores the estimation results by the normal population aggregation unit 11, and as illustrated in FIG. , non-residents, total number), the average estimated population, and the variance of the estimated population.

The disaster scale estimating unit 13 acquires population data after the occurrence of a disaster, aggregates the population at the time of the disaster for each time division in area units from the acquired population data, and calculates the population at the time of the disaster for each time division in the same area unit from the normal population in the same time division. This is a functional unit that estimates the scale of a disaster based on predetermined criteria based on changes in the population at the time of a disaster. The disaster scale estimation unit 13 in the first embodiment includes a population ratio calculation unit 13A, a population ratio table 13B, and a learning inference unit 13C.

The population ratio calculation unit 13A aggregates the population at the time of the disaster for each time segment in area (city, ward, town, village) units from the population data after the occurrence of the disaster, and calculates the population at the time of disaster relative to the normal population in the same time segment in the same area unit. This is the functional part that calculates the population ratio. As described above, the population data includes information on the estimated population for each time segment in areas (city, ward, town, village) classified by residential region, as illustrated in FIG. Therefore, the population ratio calculation unit 13A includes "residents" whose area of residence is the same as the area (city, ward, town, village), "non-residents" whose residence area is different from the area (city, ward, town, village), and both of these. For each of the "total numbers," the population at the time of the disaster is aggregated for each time period in the area (city, ward, town, village) unit, and the population ratio of the population at the time of the disaster to the normal population in the same time period in the same area unit is calculated. The results (population ratios for each time segment and population type) are stored in the population ratio table 13B.

The population ratio table 13B is a table that stores the population ratio obtained by calculation by the population ratio calculation unit 13A, and as illustrated in FIG. (Residents, non-residents, total number) and population ratio information will be stored.

The learning inference unit 13C uses the population ratio for each area and time segment obtained by the population ratio calculation unit 13A, and the label attached according to the damage level after the occurrence of a disaster obtained from the outside. ``Learning process'' (Figure 8), which performs ``machine learning'' to construct a trained estimation model, and inputs the population ratio for each area and time period after the occurrence of the disaster to be estimated into the trained estimation model. This is a functional unit that executes "inference processing" (FIG. 10) that estimates the scale of the disaster to be estimated, based on the label obtained by the calculation. Note that known techniques such as support vector machine (SVM) and neural network may be used as the machine learning method.

(Processing in the first embodiment)
An example of processing executed in the first embodiment will be described below. FIG. 5 shows the entire process in the first embodiment. As shown in FIG. 5, in normal times when no disaster has occurred, the normal population aggregation process (step S1) is executed by the normal population aggregation unit 11, and the occurrence of a disaster is used as a trigger (YES in step S2). Population ratio calculation processing (step S3) is executed by the population ratio calculation unit 13A, and disaster scale estimation processing (step S4) is executed by the learning inference unit 13C. Note that the process in Figure 5 may be executed at predetermined time intervals, or it may be executed all at once on past data at any timing (for example, when a disaster occurs, at a timing specified by a user using the system, etc.). May be executed. Each process will be explained below.

In the normal population aggregation process (step S1 in FIG. 5), as shown in FIG. From the included "area (city, ward, town, village)" and "residence area", "residents" whose area and residence area are the same, "non-residents" whose area and residence area are different, and "total number" including both of these. '', the average value and variance of the population for each time period in area (city, ward, town, village) units are estimated as the normal population (step S12), and the estimation results are stored in the normal population table 12 (step S12). S13). As a result, the stored data (average value and variance of the population for each time segment and population type in area (city, ward, town, village) unit) of the normal population table 12 illustrated in FIG. 3 is obtained.

The process in step S11 is executed at predetermined intervals (for example, at one hour intervals) unless disaster occurrence information is received from the outside, and the estimation results are stored in the normal population table 12.

When the disaster occurrence information is received from the outside, the population ratio calculation unit 13A of the disaster scale estimating unit 13 executes the following population ratio calculation process (step S3 in FIG. 5). Specifically, as shown in FIG. 7, the population ratio calculation unit 13A acquires population data after the occurrence of a disaster (step S31), and calculates the "area (city, ward, town, village)" and "residence area" included in the population data. ”, time per area (city, ward, town, village) for each of “residents” whose area and residence area are the same, “non-residents” whose area and residence area are different, and “total number” including both of these. The population at the time of disaster is totaled for each category (the average value and variance of the population are calculated as the population at the time of disaster: step S32). Furthermore, the population ratio calculation unit 13A combines the population at the time of disaster and the normal population using the city, ward, town, village, time division, and population type as keys (step S33), calculates the population ratio for each time division and each population type, It is stored in the population ratio table (step S34). As a result, the stored data (population ratio for each time segment and population type in area (city, ward, town, village) unit) of the population ratio table 13B illustrated in FIG. 4 is obtained.

Next, in the disaster scale estimation process (step S4 in FIG. 5), the learning inference unit 13C estimates the disaster scale by executing the inference process in FIG. However, in reality, in order to improve the accuracy of the inference process, the learning inference unit 13C executes the learning process in FIG. 8 every time a disaster occurs in the past, as a preparation for executing the inference process in FIG. and train a supervised machine learner. That is, the learning process and the inference process use different population data at the time of the disaster. Note that the learning process in FIG. 8 may be performed all at once on past data at any timing (for example, when a disaster occurs, at a timing specified by a user using the system, etc.).

In the learning process of FIG. 8, the learning inference unit 13C calculates the population ratio for each time period for each city, ward, town, or village (step S41). For example, if you use the population ratio every 6 hours from 12:00 instead of every hour, you will get the "population ratio for each area (city, ward, town, village), time segment, population type", which is marked as "input feature" in Figure 9. be done. Next, the learning inference unit 13C assigns a teacher label of "1" or "0" to each city, ward, town, or village depending on whether or not the Disaster Relief Act is applied (step S42). As a result, a correspondence table with a "teacher label" shown at the right end of FIG. 9 is obtained. Furthermore, the learning inference unit 13C trains the supervised machine learning device using the "population ratio as input feature" and the "teacher label as teacher data" shown in the correspondence table of FIG. 9 (step S43).

In step 4 of FIG. 5 when a disaster occurs, the learning inference unit 13C uses the same format as during learning, as shown in FIG. Population ratio data is generated (step S44). As a result, "population ratio data for each area (city, ward, town, village), for each time segment, and for each population type" after the disaster occurs, which is indicated as "input feature amount" in FIG. 11, is obtained. Next, the learning inference unit 13C inputs the generated "population ratio data for each area (city, ward, town, village), time segment, and population type" into the trained model, and uses the area (city, ward, town, village) as the estimation result. A label (“1” or “0”) for each is acquired (step S45). As a result, estimation results regarding whether or not the Disaster Relief Law is applied for each city, ward, town, or village are obtained as an example of the disaster scale for each area (city, ward, town, or village). Although absolute times such as 12:00 and 18:00 are illustrated in FIGS. 9 and 11, the time elapsed after the disaster (for example, 3 hours after the disaster occurs, 9 hours after the disaster, etc.) is used as an example. Good too.

According to the first embodiment described above, the change in population at the time of disaster (population Based on the ratio), it is possible to quickly and accurately estimate the scale of a disaster after it occurs. Accordingly, the results of estimating the scale of the disaster can be effectively utilized, for example, in disaster response by local governments.

In addition, the estimated population is not only estimated by the "total number" but also by type such as "residents" and "non-residents," and the estimation results are used as the basis for estimating the disaster scale, so disasters occur more accurately. It is possible to estimate the scale of the disaster later.

(Modified example of the first embodiment)
In the field of deep learning, it is known that estimation accuracy can be improved by pre-training an estimation model using data different from the target to be estimated, and then retraining it using the data to be estimated (Nakayama et al. (See Hideki, “Image feature extraction and transfer learning using deep convolutional neural networks,” IEICE Technical Report, vol. 115, no. 146, SP2015-45, pages 55-59, July 2015). Such pre-learning and re-learning are effective, for example, when learning data cannot be collected on a large scale.

Since the "data indicating the disaster scale" according to the present disclosure cannot be collected on a large scale as learning data, the estimation accuracy can be improved by performing "pre-learning and re-learning" as described above. For example, the first modification is "pre-learning and re-learning" using damage data of different damage levels, and the second modification is "pre-learning and re-learning" using damage data in area units with different granularity. can be mentioned.

As a specific example of Modification 1 above, if the data that you originally want to estimate is "the number of completely destroyed cases for each city, ward, town, village", the number of partially damaged cases is expected to be larger than the number of completely destroyed cases, so learning As shown in FIG. 12(a), the inference unit 13C pre-learns an estimation model using the number of partially damaged cases for each city, ward, town, and village as training data (step S46), and then uses the number of completely damaged cases for each city, ward, town, and village as training data. An example of this is relearning the estimated model as data (step S47). Estimation accuracy can be improved by performing "pre-learning and re-learning" using damage data of different damage levels (the number of partially damaged cases and the number of completely destroyed cases).

As a specific example of the above modification 2, when the data that is originally desired to be estimated is "the number of partially damaged cases per town/chome", the learning inference unit 13C estimates the "number of partially damaged cases per city, ward, town and village" as shown in FIG. 12(b). The estimation model is trained in advance using the "number of partially damaged cases in each town" as training data (step S48), and then the estimation model is retrained using the "number of partially damaged cases in each town", which is finer-grained than the city, ward, town, village, as training data. (Step S49). "Machi-chome" is finer in granularity than "city, ward, town, village" and would normally contain more data. However, data at the granularity of "town, block" (number of damage cases) is often not made public and is generally difficult to collect, whereas data at the granularity of "city, ward, town, village" (number of damage cases) is publicly available. Many are generally easy to collect. Therefore, since it is expected that the number of data that can be collected is relatively larger for the "number of partial damage cases by city, ward, town, village," Estimation accuracy can be improved by performing "pre-learning and re-learning." Note that the above-described modifications 1 and 2 may be combined and implemented.

[Second embodiment]
As shown in FIG. 13, a disaster scale estimation device 10S according to the second embodiment includes a normal population aggregation unit 11, a normal population table 12, and a disaster scale estimation unit 14. Of these, the normal population aggregation unit 11 and the normal population table 12 are the same as those in the first embodiment described above, so a redundant explanation will be omitted.

The disaster scale estimating unit 14 acquires population data after the occurrence of a disaster, aggregates the population at the time of the disaster for each time division in area units from the acquired population data, and calculates the population at the time of the disaster for each time division in the same area unit from the normal population in the same time division. This is a functional unit that estimates the scale of a disaster based on predetermined criteria based on changes in the population at the time of a disaster. The disaster scale estimation unit 14 in the second embodiment includes an abnormality score calculation unit 14A, abnormality score array data 14B, and an estimation unit 14C.

The abnormality score calculation unit 14A is a functional unit that calculates an "abnormality score", which will be described later, based on the difference between the population during a disaster and the normal population in the same time period in the same area unit, for each same time period in the same area unit. As described in the first embodiment, the population data includes information on the estimated population for each time period in area units classified by residential area as shown in FIG. The abnormality score calculation unit 14A calculates, from the information on the residential area and the area (city, town, village in this case) included in the population data, whether the residential area and the area are the same as the "residents" or the residential area and the area are different. For each of "non-residents" and "total number" including residents and non-residents, the population at the time of the disaster is aggregated for each area and time period, an abnormal score is calculated, and the calculation results are arranged in an abnormal score array. Output to data 14B.

ここで、xはある市区町村におけるある時間区分での人口、iは人口種別、tは時間区分、μは当該市区町村における当該時間区分での平均、σは当該市区町村における当該時間区分での分散を、それぞれ表す。本実施形態では、異常スコア計算部１４Ａは、災害発生からＮ時間後の時点から（Ｎ＋Ｋ）時間後の時点まで１時間間隔で（Ｋ＋１）回にわたり、「居住者」、「非居住者」、「総数」の３つの人口種別それぞれについて上記の異常スコアａを計算し、計算結果を異常スコア配列データ１４Ｂに保管する。なお、上記の式１は異常スコアの計算式の一例であり、上記の式１に限定されるものではない。また、異常スコア計算の時期に係る上記のＮ、Ｋは予め定められた定数であり、これらは適宜な値に定めてよい。例えば、Ｎ＝３、Ｋ＝３の場合は、災害発生から３時間後から６時間後まで１時間間隔で計４回にわたり上記３つの人口種別それぞれについて異常スコアａが計算される。 The "abnormality score" here is an example of an index representing the degree of change in the population at the time of disaster from the normal population, and the abnormality score calculation unit 14A calculates the abnormality score using Equation 1 below.

Here, x is the population in a certain time period in a certain city, ward, town, or village, i is the population type, t is a time period, μ is the average in that time period in the city, ward, town, or village, and σ is the corresponding time in the city, ward, town, or village. It represents the variance in each category. In the present embodiment, the abnormality score calculation unit 14A calculates "resident", "non-resident", "resident", "non-resident", The above abnormality score a is calculated for each of the three population types of "total number", and the calculation results are stored in the abnormality score array data 14B. Note that Equation 1 above is an example of a formula for calculating the abnormality score, and is not limited to Equation 1 above. Further, the above N and K related to the timing of abnormality score calculation are predetermined constants, and these may be set to appropriate values. For example, in the case of N=3 and K=3, the abnormality score a is calculated for each of the three population types a total of four times at one hour intervals from three hours after the occurrence of the disaster to six hours after the occurrence of the disaster.

The anomaly score array data 14B is a table that stores the anomaly scores a obtained by calculations by the anomaly score calculation unit 14A, and stores the anomaly scores a for each of the three population types over (K+1) times. Therefore, the number of elements in the abnormal score array data 14B is "3×(K+1)".

The estimation unit 14C is a functional unit that estimates the disaster scale according to the total number of abnormality scores that are equal to or higher than a predetermined criterion among the plurality of abnormality scores stored in the abnormality score array data 14B. The method of estimating the disaster scale by this estimator 14C will be described later.

(Processing in the second embodiment)
An example of processing executed in the second embodiment will be described below. FIG. 14 shows the entire process in the second embodiment. As shown in FIG. 14, in normal times when no disaster has occurred, the normal population aggregation process (step S1) is executed by the normal population aggregation unit 11, and the occurrence of a disaster is used as a trigger (YES in step S2). The abnormality score calculation process (step S5) is executed by the abnormality score calculation unit 14A, and the disaster scale estimation process (step S6) is executed by the estimation unit 14C. Note that the process in FIG. 14 may be executed at predetermined time intervals, or may be executed all at once on past data at any timing (for example, when a disaster occurs, at a timing specified by a user using the system, etc.). May be executed. Each process will be explained below.

The normal population aggregation process (step S1 in FIG. 14) is the same as in the first embodiment, so a duplicate explanation will be omitted. Through this estimation process, the stored data (average value and variance of population for each time segment and population type in area (city, ward, town, village) unit) of the normal population table 12 illustrated in FIG. 3 is acquired. The process of step S1 is executed at predetermined intervals (for example, at one hour intervals) unless disaster occurrence information is received from the outside, and the estimation results are stored in the normal population table 12.

When disaster occurrence information is received from the outside, the abnormality score calculation unit 14A of the disaster scale estimating unit 14 executes the following abnormality score calculation process (step S5 in FIG. 14). Specifically, as shown in FIG. 15, the abnormality score calculation unit 14A calculates the population average and variance at the time of disaster for each city, ward, town, and village by population type, and also calculates the same time segment for each city, ward, town, and village by population type. The normal population average and variance of are obtained from the normal population table 12 (step S51). Then, the abnormality score calculation unit 14A calculates the abnormality score for the same time period for each city, ward, town, village and population type from the "population average and variance during disaster" and "population average and variance during normal times" acquired in step S51. is calculated using Equation 1 described above and stored in the abnormal score array data 14B (step S52). Here, as described above, the abnormality score calculation unit 14A calculates the number of "residents" and "non-residents" for a total of (K+1) times from the time N hours after the disaster occurrence to the time (N+K) hours after the occurrence of the disaster. , "total number" is calculated for each of the three population types, and the calculation results are stored in the anomaly score array data 14B. As a result, 3×(K+1) abnormal scores are stored in the abnormal score array data 14B.

Next, in the disaster scale estimation process (step S6 in FIG. 14), the estimation unit 14C estimates the disaster scale by executing the process in FIG. 16 described below. Specifically, the estimation unit 14C determines whether each of the 3×(K+1) abnormality scores stored in the abnormality score array data 14B is equal to or greater than a predetermined threshold, and as a result, the threshold It is determined whether the total number of abnormality scores is equal to or greater than a predetermined reference number (T) (step S61). The reference number T here is a value predetermined as a standard for estimating whether the damage is small (the scale of the disaster is small) or the damage is large (the scale of the disaster is large).

If the total number of abnormality scores that are equal to or greater than the threshold is T or more in step S61, the estimating unit 14C outputs the estimation result that "damage is large (disaster scale is large)" (step S62), and ends the process. do. On the other hand, if the total number of anomaly scores that are equal to or greater than the threshold is not equal to or greater than T in step S61, the estimation unit 14C outputs the estimation result that "damage is small (disaster scale is small)" (step S63), and processes end.

According to the second embodiment described above, by effectively utilizing the fact that accurate population data can be obtained quickly from the outside, abnormalities based on the difference between the population at the time of disaster and the population during normal times in the same time period in the same area unit are detected. Based on the score, it is possible to quickly and accurately estimate the scale of a disaster after it occurs. Accordingly, the results of estimating the scale of the disaster can be effectively utilized, for example, in disaster response by local governments. In addition, the estimated population is not only estimated by the "total number" but also by type such as "residents" and "non-residents," and the estimation results are used as the basis for estimating the disaster scale, so disasters occur more accurately. It is possible to estimate the scale of the disaster later.

In addition, in the process of FIG. 16, an example was shown in which the scale of the disaster is estimated by binary classification of damage being large/small. By defining multiple classifications, it is possible to estimate the disaster scale using three or more classifications.

In addition, in the process of FIG. 16, an example was shown in which 3×(K+1) abnormality scores are evaluated at the same level, but for example, the abnormality score for residents, the abnormality score for non-residents, and the total number of abnormality scores are evaluated at the same level. A total of three types of abnormality scores may be weighted and evaluated as appropriate. In addition, (K+1) types of anomaly score N hours after the disaster occurrence, anomaly score after (N+1) hours, ..., anomaly score after (N+K) hours (that is, according to the elapsed time after the disaster occurrence) (type) may be weighted appropriately for evaluation.

In the first and second embodiments described above, the estimated population is estimated not only by the "total number" but also by type such as "residents" and "non-residents," and the estimation results are used as the basis for estimating the disaster scale. However, it is not necessary to estimate "residents" and "non-residents," and in order to simplify the process, the estimation results of only the "total number" are used as the basis for estimating the disaster scale. Good too.

[About terminology, deformation, etc.]
It should be noted that the block diagram used to explain the above embodiment shows blocks in functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method for realizing each functional block is not particularly limited. That is, each functional block may be realized using one physically or logically coupled device, or may be realized using two or more physically or logically separated devices directly or indirectly (e.g. , wired, wireless, etc.) and may be realized using a plurality of these devices. The functional block may be realized by combining software with the one device or the plurality of devices.

Functions include judgment, decision, judgment, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, These include, but are not limited to, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assigning. I can't do it. For example, a functional block (configuration unit) that performs transmission is called a transmitting unit or transmitter. In either case, as described above, the implementation method is not particularly limited.

For example, the disaster scale estimating device in one embodiment may function as a computer that performs the processing in this embodiment. FIG. 17 is a diagram showing an example of the hardware configuration of the disaster scale estimating device 10. The disaster scale estimating device 10 described above may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like. Note that the disaster scale estimating device 10S in FIG. 13 may also have the same configuration as the disaster scale estimating device 10 described below.

In addition, in the following description, the word "apparatus" can be read as a circuit, a device, a unit, etc. The hardware configuration of the disaster scale estimating device 10 may be configured to include one or more of each device shown in the figure, or may be configured not to include some of the devices.

Each function in the disaster scale estimating device 10 is such that the processor 1001 performs calculations by loading predetermined software (programs) onto hardware such as the processor 1001 and the memory 1002, and controls communication by the communication device 1004. This is realized by controlling at least one of reading and writing data in the memory 1002 and storage 1003.

The processor 1001, for example, operates an operating system to control the entire computer. The processor 1001 may be configured by a central processing unit (CPU) including an interface with a peripheral device, a control device, an arithmetic device, a register, and the like.

Furthermore, the processor 1001 reads programs (program codes), software modules, data, and the like from at least one of the storage 1003 and the communication device 1004 to the memory 1002, and executes various processes in accordance with these. As the program, a program that causes a computer to execute at least part of the operations described in the above embodiments is used. Although the various processes described above have been described as being executed by one processor 1001, they may be executed by two or more processors 1001 simultaneously or sequentially. Processor 1001 may be implemented by one or more chips. Note that the program may be transmitted from a network via a telecommunications line.

The memory 1002 is a computer-readable recording medium, and includes at least one of ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), and the like. may be done. Memory 1002 may be called a register, cache, main memory, or the like. The memory 1002 can store executable programs (program codes), software modules, and the like to implement a wireless communication method according to an embodiment of the present disclosure.

The storage 1003 is a computer-readable recording medium, such as an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (such as a compact disk, a digital versatile disk, or a Blu-ray disk). (registered trademark disk), smart card, flash memory (eg, card, stick, key drive), floppy disk, magnetic strip, etc. Storage 1003 may also be called an auxiliary storage device. The storage medium mentioned above may be, for example, a database including at least one of memory 1002 and storage 1003, a server, or other suitable medium.

The communication device 1004 is hardware (transmission/reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, or the like.

The input device 1005 is an input device (eg, keyboard, mouse, microphone, switch, button, sensor, etc.) that accepts input from the outside. The output device 1006 is an output device (for example, a display, a speaker, an LED lamp, etc.) that performs output to the outside. Note that the input device 1005 and the output device 1006 may have an integrated configuration (for example, a touch panel). Further, each device such as the processor 1001 and the memory 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses for each device.

Each aspect/embodiment described in this disclosure may be used alone, may be used in combination, or may be switched and used in accordance with execution. In addition, notification of prescribed information (for example, notification of "X") is not limited to being done explicitly, but may also be done implicitly (for example, not notifying the prescribed information). Good too.

Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described in the present disclosure. The present disclosure can be implemented as modifications and variations without departing from the spirit and scope of the present disclosure as determined by the claims. Therefore, the description of the present disclosure is for the purpose of illustrative explanation and is not intended to have any limiting meaning on the present disclosure.

The order of the processing procedures, sequences, flowcharts, etc. of each aspect/embodiment described in this disclosure may be changed as long as there is no contradiction. For example, the methods described in this disclosure use an example order to present elements of the various steps and are not limited to the particular order presented.

The input/output information may be stored in a specific location (eg, memory) or may be managed using a management table. Information etc. to be input/output may be overwritten, updated, or additionally written. The output information etc. may be deleted. The input information etc. may be transmitted to other devices.

As used in this disclosure, the phrase "based on" does not mean "based solely on" unless explicitly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

Where "include", "including" and variations thereof are used in this disclosure, these terms, like the term "comprising," are inclusive. It is intended that Furthermore, the term "or" as used in this disclosure is not intended to be exclusive or.

In this disclosure, when articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are plural.

In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." Note that the term may also mean that "A and B are each different from C". Terms such as "separate" and "coupled" may also be interpreted similarly to "different."

10, 10S... Disaster scale estimating device, 11... Normal population counting unit, 12... Normal population table, 13... Disaster scale estimation unit, 13A... Population ratio calculation unit, 13B... Population ratio table, 13C... Learning inference unit, 14... Disaster scale estimation unit, 14A... Abnormality score calculation unit, 14B... Abnormality score array data, 14C... Estimation unit, 1001... Processor, 1002... Memory, 1003... Storage, 1004... Communication device, 1005... Input device, 1006... Output device, 1007...bus.

Claims (7)

Population data that includes information on estimated population for each time segment in a predetermined area unit, wherein the population data in normal times is obtained, and the normal population for each time segment in the area unit is obtained from the obtained population data. a normal population counting department that tabulates the
Obtain the population data after the occurrence of a disaster, aggregate the population at the time of the disaster for each time segment in the area unit from the obtained population data, and calculate the population at the time of the disaster from the normal population in the same time segment in the same area unit. a disaster scale estimating unit that estimates the disaster scale based on population changes and in light of predetermined criteria;
Equipped with
The population data includes information on the estimated population for each time segment in the area unit further classified by residential area based on the user's address attribute,
The normal population counting unit calculates the normal population for each time period in the area unit for each of residents whose residential area and the area are the same and non-residents whose residential area and the area are not the same. Calculate the population at the time,
The disaster scale estimating unit estimates the disaster scale by totaling the population at the time of the disaster for each time period of the area unit, for each of the residents and non-residents.
Disaster scale estimation device. The disaster scale estimating unit estimates the disaster scale based on the ratio of the disaster population to the normal population in the same area unit and in the same time period.
The disaster scale estimating device according to claim 1 . The disaster scale estimation department is
Obtain the population data after the occurrence of the disaster, aggregate the population at the time of disaster for each time segment in the area unit from the obtained population data, and calculate the disaster population with respect to the normal population in the same time segment in the same area unit. a population ratio calculation unit that calculates the population ratio of the current population;
Perform supervised machine learning using the population ratio for each time period in the area unit obtained by the population ratio calculation unit and the label attached according to the damage level after the disaster occurs to create a learned estimation model. At the same time, the disaster scale of the disaster to be estimated is estimated based on the label obtained by inputting the population ratio for each time period of the area unit after the occurrence of the disaster to be estimated into the trained estimation model. A learning inference unit that performs
The disaster scale estimating device according to claim 2 , comprising: The learning inference unit is
Pre-learning a trained estimation model by performing supervised machine learning using a label attached according to the first damage level and the population ratio,
The pre-trained trained estimation model is retrained by performing supervised machine learning using the population ratio and a label attached according to a second damage level different from the first damage level. do,
The disaster scale estimating device according to claim 3 . The learning inference unit is
pre-learning a trained estimation model by performing supervised machine learning using the population ratio for each first area unit and a label attached according to the damage level;
By performing supervised machine learning using the population ratio for each second area unit having a different granularity from the first area unit and a label attached according to the damage level, Retrain the trained estimation model,
The disaster scale estimating device according to claim 3 or 4 . The disaster scale estimating unit estimates the disaster scale based on the difference between the disaster population and the normal population in the same area unit and in the same time period.
The disaster scale estimating device according to claim 1 . The disaster scale estimation department is
an abnormality score calculation unit that calculates an abnormality score based on the difference between the population at the time of disaster and the normal population in the same area unit and the same time period, for each same time period in the same area unit;
an estimation unit that estimates a disaster scale according to the total number of abnormality scores that are equal to or higher than a predetermined criterion among the plurality of abnormality scores obtained by the calculation by the abnormality score calculation unit;
The disaster scale estimating device according to claim 6 , comprising:

Images