Prediction device

Abstract

This prediction device comprises a storage unit that stores at least one prediction model, an input unit that receives information about the type of a target disaster and information about a target cause, and a control unit. The target disaster is a disaster that a first structure, for which a disaster damage status is to be predicted, is predicted to undergo. The target cause is a cause of damage that the first structure is predicted to receive when the first structure undergoes the target disaster. The prediction model is generated from data pertaining to a second structure different from the first structure. The control unit calls, from the storage unit, a prediction model that is associated with a disaster of the same type as the type of the target disaster and is associated with the same cause of damage as the target cause. The control unit acquires a result of prediction of the damage status in the first structure on the basis of the called prediction model.

Description

Prediction Device

[0001] The present disclosure relates to a prediction device.

[0002] When a disaster occurs, structures may be damaged, so it is important to predict the damage to structures when a disaster occurs.

[0003] Here, a prediction model is used to predict the damage status of a structure. This prediction model is constructed using damage data from when a structure of the same type as the structure whose damage status is to be predicted actually suffered damage. As an example of constructing this prediction model, Non-Patent Document 1 describes constructing an estimation formula using damage results for water pipes damaged in an earthquake. Furthermore, Non-Patent Document 2 describes constructing a prediction model for communication pipes using damage data for underground pipes damaged in an earthquake.

[0004] Ryuji Isoyama et al., "Study on Earthquake Damage Prediction of Water Pipelines," Journal of Japan Water Works Association, No. 67 (2), February 1998, pp. 36-51. Yo Ito et al., "Study on Damage Prediction Models Using Machine Learning Based on Inspection Results of Underground Telecommunication Pipelines During Earthquakes," Proceedings of the Japan Society of Civil Engineers, A1 (Structural and Earthquake Engineering), Vol. 78, No. 4, 2022, pp. I_162-I_172.

[0005] However, there are some structures for which it is difficult to obtain sufficient damage data in advance, making it difficult to build a predictive model for such structures.

[0006] The purpose of the present disclosure, made in consideration of the above, is to predict the damage situation of structures for which sufficient damage data has not been acquired.

[0007] A prediction device according to one embodiment of the present disclosure comprises: a memory unit that stores at least one prediction model, wherein the prediction model is linked to information on the type of disaster and information on damage factors; an input unit that accepts information on the type of target disaster and information on target factors; and a control unit, wherein the target disaster is a disaster that is predicted to occur to a first structure whose damage situation is to be predicted; the target factors are factors of damage that are predicted to occur when the first structure is hit by the target disaster; the prediction model is generated from data on a second structure that is different from the first structure; and the control unit calls up from the memory unit the prediction model that is linked to the disaster of the same type as the type of target disaster and to the same damage factors as the target factors, and obtains a prediction result for the damage situation of the first structure based on the called prediction model.

[0008] According to one embodiment of the present disclosure, it is possible to predict the damage status of a structure for which sufficient damage data has not been acquired.

[0009] FIG. 2 is a block diagram illustrating an example configuration of a prediction device according to an embodiment of the present disclosure. FIG. 3 is a flowchart illustrating an example procedure of the prediction device shown in FIG. 1. FIG. 4 is a diagram illustrating an example processing of the prediction device shown in FIG. 1. FIG. 5 is a diagram illustrating an example processing of the prediction device shown in FIG. 1. FIG. 6 is a diagram illustrating an example processing of the prediction device shown in FIG. 1. FIG. 7 is a diagram illustrating an example processing of the prediction device shown in FIG. 1. FIG. 8 is a diagram illustrating an example processing of a third modified example of the prediction device shown in FIG. 1. FIG. 9 is a diagram illustrating an example processing of the third modified example of the prediction device shown in FIG. 1. FIG. 10 is a diagram illustrating damage data for generating a prediction model.

[0010] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0011] (Configuration of Prediction Device) A prediction device 10 according to this embodiment, as shown in Fig. 1, can predict the damage status of a first structure 2 when a disaster occurs, using a prediction model 1. The prediction model 1 is generated using data on a second structure 3 that is separate from the first structure 2. In other words, the prediction device 10 can predict the damage status of the first structure 2 even if there is no damage data from when the first structure 2 was actually damaged. In this embodiment, the prediction device 10 acquires a damage rate of the first structure 2 as a prediction result of the damage status of the first structure 2, as will be described later.

[0012] The first structure 2 and the second structure 3 are, for example, utility poles, roads, communication pipes, steel towers, mountain bodies, or any other infrastructure facilities. However, the first structure 2 and the second structure 3 are not limited to these. The first structure 2 and the second structure 3 may be any structure as long as they are structures.

[0013] The prediction device 10 is, for example, an information processing device such as a desktop personal computer (PC) or a notebook PC.

[0014] The prediction device 10 includes an input unit 11 , a display unit 12 , a storage unit 13 , and a control unit 16 .

[0015] The input unit 11 can receive input from a user. The input unit 11 includes at least one input interface that can receive input from a user. The input interface is, for example, a physical key, a capacitance key, a pointing device, a touch screen that is integrated with the display of the display unit 12, a microphone, or the like.

[0016] The input unit 11 receives data on the first structure 2, the target of which damage status is to be predicted, from the user. The data on the first structure 2 input from the input unit 11 includes information on the type of the first structure 2, information on the type of target disaster, information on the target cause, and shape data of the first structure 2.

[0017] The type information of the first structure 2 is, for example, information about a utility pole, a road, a communication pipe, a steel tower, or a mountain. For example, suppose the user wants to predict the damage situation when a road is damaged. In this case, the type information of the first structure 2 input from the input unit 11 is information about a road.

[0018] The target disaster is a disaster that is predicted to hit the first structure 2. In other words, the prediction device 10 predicts the damage state of the first structure 2 when the first structure 2 is hit by the target disaster. For example, assume that the first structure 2 is a road, and the user wants to predict the damage state when the road is hit by a disaster caused by heavy rain. In this case, the target disaster input from the input unit 11 is information about heavy rain.

[0019] The target cause is a cause of target damage that is predicted to occur when the first structure 2 is hit by a target disaster. For example, if the first structure 2 is a road, it is predicted that the road will suffer damage from road closure when the road is hit by a target disaster such as heavy rain. Furthermore, the causes of road closure, i.e., the causes of damage from road closure, are fallen trees and road collapse. In this case, the target causes input from the input unit 11 are fallen trees and road collapse.

[0020] The shape data of the first structure 2 is data that indicates the shape of the first structure 2. Any point in the shape data of the first structure 2 may be associated with position information and altitude information.

[0021] The display unit 12 is capable of displaying images. The display unit 12 includes a display. The display is, for example, an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) display.

[0022] The memory unit 13 is configured to include at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The memory unit 13 may function as a main memory, an auxiliary memory, or a cache memory. The memory unit 13 stores data used in the operation of the prediction device 10 and data obtained by the operation of the prediction device 10. The memory unit 13 may store a program executed by the control unit 16. The memory unit 13 may store information on the type of the first structure 2, information on the target disaster, information on the target cause, and information on damage to the first structure 2 for which a damage rate is calculated, linked to each other. The information on damage to the first structure 2 for which a damage rate is calculated may include information on the type of damage and information on the extent of the damage.

[0023] The storage unit 13 stores at least one prediction model 1. In this embodiment, the storage unit 13 stores a data group 14. The data group 14 includes a plurality of data 15. The data 15 includes the prediction model 1, information on the type of second structure 3, information on the type of disaster 4, and information on the damage factor 5. In the data 15, the prediction model 1, the information on the type of second structure 3, the information on the type of disaster 4, and the information on the damage factor 5 are linked together.

[0024] The prediction model 1 is a model of a plurality of meshes M ｉ (i is an integer satisfying 2≦i≦N) and each mesh M ｉ Damage rate R of the second structure 3 ｉ and data.

[0025] Mesh M ｉ The mesh M is associated with position information and altitude information. ｉ is, for example, a two-dimensional mesh, where mesh M ｉ may be a three-dimensional mesh. ｉ For example, the mesh M given by the Statistics Bureau of the Ministry of Internal Affairs and Communications (https: / / www.stat.go.jp / data / mesh / m_tuite.html) may be used. ｉThe mesh M may be given by any model such as the Digital Elevation Model (DEM) of the Geospatial Information Authority of Japan (https: / / fgd.gsi.go.jp / download / ref_dem.html). ｉ The size is expressed as, for example, length [m] x width [m].

[0026] Damage rate R ｉ is mesh M ｉ is the probability that the second structure 3 will be damaged by the damage factor 5 when a disaster 4 occurs. Here, the damage rate R ｉ is expressed as in equation (1). ｉ = f(X) Equation (1) In equation (1), X = (X １ , X ２ , …, X Ｌ ) The variable X １ ~X Ｌ (L is an integer satisfying 1≦L) is an explanatory variable. １ ~X Ｌ is set by the setting environment or information of the second structure 3. １ ~X Ｌ For example, if the second structure 3 is a utility pole, the information of the second structure 3 set in the variable X １ ~X Ｌ For example, if the second structure 3 is a communication line, the information of the second structure 3 set in the variable X is information on the type of pipe of the communication line. １ ~X Ｌ Any of the meshes M ｉ corresponds to the size of

[0027] The information on the type of the second structure 3 is, for example, information on a utility pole, a road, a communication pipe, a steel tower, or a mountain.

[0028] The type of disaster 4 is the type of disaster that has actually occurred to the second structure 3. The type of disaster 4 is, for example, heavy rain, earthquake, volcano, or windstorm.

[0029] The information on damage cause 5 is information on the cause of the damage that the second structure 3 actually suffered when it was hit by disaster 4. As an example, it is assumed that the second structure 3 is a utility pole, and that the utility pole suffered damage such as collapse or tilting when it was hit by disaster 4. It is also assumed that the causes of the utility pole collapsing or tilting were fallen trees and a road collapse. In this case, damage cause 5 is fallen trees and a road collapse. As another example, it is assumed that the second structure 3 is a road, and that the road suffered damage such as road closure when it was hit by disaster 4. It is also assumed that the causes of the road becoming blocked are fallen trees and a road collapse. In this case, damage cause 5 is fallen trees and a road collapse.

[0030] In FIG. 1, the plurality of data 15 includes data 15a, 15b, 15c, and 15c.

[0031] The data 15a includes information on the type of the second structure 3A, information on the type of the disaster 4a, and information on the damage cause 5a. The type of the second structure 3A is a utility pole. The type of the disaster 4a is heavy rain. The damage cause 5a is a fallen tree and a collapsed road.

[0032] The data 15b includes information on the type of the second structure 3A, information on the type of disaster 4a, and information on the damage cause 5b. The damage cause 5b is external flooding.

[0033] The data 15c includes information on the type of the second structure 3A, information on the type of the disaster 4b, and information on the cause of damage 5c. The cause of damage 5c is a fire.

[0034] The data 15d includes information on the type of the second structure 3B, information on the type of the disaster 4b, and information on the cause of damage 5a. The type of the second structure 3B is a road.

[0035] The control unit 16 is configured to include at least one processor, at least one dedicated circuit, or a combination of these. The processor is, for example, a general-purpose processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit), or a dedicated processor specialized for a specific process. The dedicated circuit is, for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). The control unit 16 executes processes related to the operation of the prediction device 10 while controlling each part of the prediction device 10.

[0036] (Operation of Prediction Device) FIG. 2 is a flowchart showing an example of a procedure of the prediction device 10 shown in FIG.

[0037] The control unit 16 receives data on the first structure 2 from the user via the input unit 11 (step S1). As described above, the data on the first structure 2 input from the input unit 11 includes information on the type of the first structure 2, information on the type of target disaster, information on the target cause, and shape data of the first structure 2.

[0038] The control unit 16 searches the data group 14 and calls up a prediction model 1 that is linked to the same type of disaster as the target disaster type received in the processing of step S1 and that is linked to the same damage factor as the target factor received in the processing of step S1 (step S2).

[0039] For example, assume that the type of target disaster received in step S1 is heavy rain and the target causes received in step S1 are fallen trees and road collapses. In this case, in step S2, the control unit 16 searches the data group 14 and calls the prediction model 1a.

[0040] Here, in the processing of step S2, the type of the second structure 3 linked to the prediction model 1 to be called may be different from the type of the first structure 2. In other words, even if the type of the first structure 2 and the type of the second structure 3 are different, the control unit 16 may call up the prediction model 1 as long as it is linked to the same type of disaster as the type of target disaster and to the same damage factor as the target factor.

[0041] The control unit 16 virtually places the second structure 3 of the prediction model 1 called in the processing of step S2 on the first structure 2 (step S3). In this embodiment, the control unit 16 virtually places the second structure 3 on the first structure 2 by placing an object of the second structure 3 in the shape data of the first structure 2. The control unit 16 virtually places the second structure 3 on the first structure 2 by placing an object of the second structure 3 on the shape data of the first structure 2. ｉ For each mesh M associated with altitude information of 10 [m], an object of the second structure 3 may be placed on the shape data of the first structure 2. ｉ When the size of the first structure 2 is the smallest, the control unit 16 creates a 10 [m] × 10 [m] mesh M ｉ For example, the object of the second structure 3 is arranged in each mesh M ｉ When the size of the first structure 2 is the smallest, the control unit 16 creates a mesh M of 250 [m] x 250 [m] on the shape data of the first structure 2. ｉ An object of the second structure 3 is placed at each location.

[0042] For example, assume that the shape data of the first structure 2 is the shape data 20 shown on the left side of Fig. 3A. In Fig. 3A, the type of the first structure 2 is a road. The type of the second structure 3 is a utility pole. In Fig. 3A, the smallest unit mesh M constituting the prediction model 1 of the second structure 3 is ｉ The size of the mesh M is 10 [m] x 10 [m], which is associated with the altitude information of 10 [m]. In this case, as shown on the right side of FIG. 3A, the control unit 16 creates a 10 [m] x 10 [m] mesh M ｉ For each of the above steps, an object 21 of the second structure 3 is virtually placed on the shape data 20 .

[0043] The control unit 16 applies the prediction model 1 to the second structure 3 virtually placed on the first structure 2 in the process of step S3 (step S4). In this embodiment, the control unit 16 calculates the damage rate R ｉ Variable X of １ ~X Ｌ The control unit 16 applies the prediction model 1 to the second structure 3 by substituting a value based on the environmental settings of the first structure 2 into the variable X １ ~X Ｌ Each mesh M after a value based on the environmental setting of the first structure 2 is substituted ｉ Damage rate R of the second structure 3 ｉ For example, if the second structure 3 is a utility pole, and the variable X １ is the value of the altitude at which the utility pole is installed. In this case, the control unit 16 １ The elevation value of the second structure 3, which is a utility pole when the second structure 3 is virtually placed on the first structure 2, is substituted into .

[0044] In the process of step S4, the control unit 16 ｉ In equation (1), the damage rate R ｉ Variable X of １ ~X Ｌ For example, when the second structure 3 is a utility pole, the control unit 16 assigns the same information to all the meshes M ｉ In the variable X, the information on the structure of the utility pole is substituted. １ ~X Ｌ For example, when the second structure 3 is a communication pipe, the control unit 16 substitutes information about the same utility pole structure into all the meshes M ｉ In the variable X, information on the type of pipe for the communication pipe is substituted. １ ~X Ｌ All of the above are substituted with information about the same pipe type.

[0045] For example, FIG. 3B shows a configuration in which the prediction model 1 is applied to a second structure 3 virtually placed on a first structure 2. In FIG. 3B, the damage rate R ｉ Variable X of １ ~X ＮBy substituting a value based on the environmental settings of the first structure 2 into ｉ The damage rate R of the second structure 3 ｉ In FIG. 3B , among the objects 21 of the second structure 3, the objects 21 with a high damage rate are colored darker, and the objects 21 with a low damage rate are colored lighter.

[0046] The control unit 16 sets a calculation range for calculating the damage rate of the first structure 2 in the shape data of the first structure 2 (step S5). The calculation range may be set based on the type of damage predicted to be sustained by the first structure 2. Alternatively, the calculation range may be set based on the range of damage predicted to be sustained by the first structure 2. Here, as a method for acquiring the type of damage predicted to be sustained by the first structure 2 and the range of damage, as described above, the storage unit 13 may store information on the type of the first structure 2, information on the target disaster, information on the target cause, and information on the damage of the first structure 2 for which the damage rate is to be calculated, in association with each other. As described above, the information on the damage of the first structure 2 for which the damage rate is to be calculated may include information on the type of damage and information on the range of damage. In this case, the control unit 16 may acquire, from the storage unit 13, information on the type of damage or information on the range of damage, which is associated with the information on the type of the first structure 2, information on the type of the target disaster, and information on the target cause received in the processing of step S1, depending on the processing content of step S5.

[0047] For example, in Fig. 3C, the first structure 2, which is a road, is predicted to suffer damage from road closure when the target causes are fallen trees and road collapse. Therefore, in Fig. 3C, a calculation range for calculating the damage rate from road closure is set in the shape data 20 of the first structure 2. In Fig. 3C, the control unit 16 sets the calculation range for each L [m] in the shape data 20 of the first structure 2, as shown by the dashed lines.

[0048] The control unit 16 calculates the damage rate of the first structure 2 for each calculation range set in the process of step S5 (step S6). ｉ Damage rate R ｉFor each calculation range set in the process of step S5, the control unit 16 calculates the damage rate of the first structure 2 for each calculation range. ｉ Damage rate R ｉ The maximum value among the plurality of meshes M included in the calculation range may be used as the damage rate of the first structure 2 in the calculation range. ｉ Damage rate R ｉ As another example of aggregation, the control unit 16 may calculate the damage rate of the first structure 2 in the calculation range by calculating the average value of the damage rate of the first structure 2 in the calculation range. ｉ Damage rate R ｉ The minimum value among these may be used as the damage rate of the first structure 2 in the calculation range.

[0049] The control unit 16 calculates the calculation range C of the first structure 2 using, for example, equation (2). ｊ Damage rate R ｊ ＊ Calculate R ｊ ＊ = F (R Ｃｊ ) Equation (2) In equation (2), the damage rate R Ｃｊ is the damage rate R in Equation (1). ｉ Of these, calculation range C ｊ Mesh M ｉ Damage rate R that includes ｉ In the formula (2), the function F is ｊ Multiple damage rates R Ｃｊ For example, if multiple damage rates R Ｃｊ If the maximum value of these is set as the damage rate of the first structure 2, then F=MAX. In this case, formula (2) becomes formula (3).

[0050] R ｊ ＊ =MAX(R Ｃｊ ) Formula (3)

[0051] For example, in Fig. 3D , the control unit 16 calculates the damage rate of the road blockage caused by the first structure 2 for each calculation range indicated by the dashed line. In Fig. 3D , similar to or identical to Figs. 3B and 3C , areas with high damage rates are colored darker, and areas with low damage rates are colored lighter.

[0052] The control unit 16 causes the display unit 12 to display the damage rate data of the first structure 2 calculated in the processing of step S6 (step S7). The control unit 16 may assign the damage rate data of the first structure 2 to map data and display it on the display unit 12, or may assign the damage rate data of the first structure 2 to equipment data indicating the structure of the first structure 2 and display it on the display unit 12. For example, as shown in FIG. 3D , the control unit 16 may cause the display unit 12 to display the shape data 20 of the first structure 2 to which the damage rate of the first structure 2 has been assigned.

[0053] <First Modification> The processes of steps S2 to S4 according to the first modification will be described below.

[0054] As described above, in the processing of step S2, the control unit 16 calls a prediction model 1. That is, it has been described that the control unit 16 searches the data group 14 and calls a prediction model 1 that is linked to the same type of disaster as the type of target disaster received in the processing of step S1 and that is linked to the same damage factor as the target factor received in the processing of step S1. Here, the data group 14 may include multiple different prediction models 1 that are linked to the same type of disaster as the type of target disaster and that are linked to the same damage factor as the target factor. In this case, in the processing of step S2 according to the first modification, the control unit 16 calls multiple different prediction models 1 that are linked to the same type of disaster as the type of target disaster and that are linked to the same damage factor as the target factor.

[0055] For example, assume that the first structure 2 is a road, the target disaster is heavy rain, the target causes are fallen trees and road collapses, and the probability of road closure is calculated as the damage rate of the first structure 2. In this case, the data group 14 includes data 15e in addition to data 15a to 15d as shown in FIG. 1 . Data 15e includes prediction model 1e, information on the type of second structure 3C associated with the building, information on disaster 4a associated with heavy rain, and information on damage factors 5a associated with fallen trees and road collapses. In this case, in the processing of step S2 according to the first modification, the control unit 16 calls prediction model 1a and prediction model 1e, which are linked to the same disaster 4a as the target disaster associated with heavy rain and the same damage factor 5a as the target cause associated with fallen trees and road collapses.

[0056] In the processing of step 3 according to the first modified example, the control unit 16 virtually places a plurality of second structures 3 on the first structure 2 .

[0057] In the process of step S4 according to the first modified example, the control unit 16 applies a plurality of different prediction models 1 to a plurality of second structures 3 virtually arranged on the first structure 2. In the process of step S4 according to the first modified example, the control unit 16 calculates the damage rates R of the plurality of second structures 3 as a result of applying the plurality of different prediction models 1. ｉ Here, in the process of step S4 according to the first modified example, the control unit 16 obtains the damage rate R ｉ Based on this, the mesh M ｉ The damage rate R of one representative ｉ r or one representative damage rate Rr is acquired. An example of this process will be described below.

[0058] Damage rate R of one representative ｉ As an example of the process of acquiring r, the mesh M of the plurality of different prediction models 1 called in the process of step S1 is ｉ In this case, the control unit 16 controls the size of each mesh M of the plurality of second structures 3. ｉ Damage rate R ｉ The maximum value of the representative damage rate R ｉ It may be taken as r.

[0059] Damage rate R of one representativeｉ As another example of the process of acquiring r, the mesh M of the plurality of different prediction models 1 called in the process of step S1 can be obtained. ｉ In this case, the control unit 16 calculates the damage rates R ｉ The representative damage rate R is calculated by multiplying and adding up the weighting coefficients w set for each of the plurality of second structures 3. ｉ The weighting coefficient w may be set based on the structure of the second structure 3. For example, in the above example, the damage rate R obtained as a result of applying the second structure 3A to the prediction model 1a is ｉ is the damage rate R ｉｐ and the weighting coefficient w set for the second structure 3A is the weighting coefficient W ｐ In addition, the damage rate R obtained as a result of applying the second structure 3C to the prediction model 1e is ｉ is the damage rate R ｉｂ and the weighting coefficient w set for the second structure 3C is the weighting coefficient W ｂ In this case, the representative damage rate R ｉ r is given by equation (4). ｉ r = R ｉｐ ×W ｐ +R ｉｂ ×W ｂ Formula (4)

[0060] As an example of the process of acquiring one representative damage rate Rr, the mesh M of the plurality of different prediction models 1 called in the process of step S1 is ｉ In this case, the control unit 16 controls the size of all the meshes M ｉ Damage rate R ｉ The maximum value among these may be acquired as the representative damage rate Rr.

[0061] As another example of the process of acquiring one representative damage rate R, the mesh M of the plurality of different prediction models 1 called in the process of step S1 is used. ｉ The size of the mesh M may vary depending on the prediction model 1. In this case, the control unit 16 may divide the mesh M ｉ Damage rate R ｉThe control unit 16 may calculate the average value of the damage rate R ｉ The representative damage rate Rr may be obtained by multiplying the average value of the damage rates R and the weighting coefficient w set for each of the second structures 3 and adding up the multiplied values. The weighting coefficient w may be set based on the structure of the second structures 3, etc. For example, in the above example, the damage rate R obtained as a result of applying the second structure 3A to the prediction model 1a may be ｉ is the damage rate R ｉｐ and the weighting coefficient w set for the second structure 3A is the weighting coefficient W ｐ In addition, the mesh M obtained as a result of applying the second structure 3C to the prediction model 1e with i=k is ｋ Damage rate R ｋ is the damage rate R ｋｂ and the weighting coefficient w set for the second structure 3C is the weighting coefficient W ｂ The mesh M ｉ Size and mesh size ｋ In this case, the representative damage rate Rr is given by equation (5). In equation (5), the integer Na is the mesh M of the prediction model 1a. ｉ The integer Nb is the total number of meshes M ｋ The total number of

[0062] In the first modified example, the processes of steps S6 to S8 are the same as or similar to the processes of steps S6 to S8 described above. However, in the first modified example, the control unit 16 calculates the damage rate R ｉ As a representative damage rate R ｉ r or damage rate Rr is used.

[0063] <Second Modification> The processes of steps S2 to S4 according to the second modification will be described below.

[0064] As described above, in the processing of step S1, the control unit 16 is described as receiving data of the first structure 2 from the user via the input unit 11. Here, multiple target factors may be predicted for one target disaster predicted to affect the first structure 2. In this case, the data of the first structure 2 received in the processing of step S1 includes multiple different target factors.

[0065] In the processing of step S2 according to the second modification, when the control unit 16 receives multiple target factors for one target disaster in the processing of step S1, the control unit 16 may call multiple different prediction models 1. The multiple different prediction models 1 to be called are linked to the same type of disaster as the type of the target disaster and to the same damage factor as at least one of the multiple different target factors. However, the damage factors linked to the multiple different prediction models 1 to be called are different from each other.

[0066] For example, if the first structure 2 is a road and the road is affected by an earthquake, the user may wish to predict the damage to the road due to earthquake vibrations and liquefaction. In this case, in step S1, the user inputs the target disaster associated with an earthquake, the target factor associated with vibrations, and the target factor associated with liquefaction via the input unit 11. Here, the data group 14 includes data 15f and 15g in addition to data 15a-15d shown in FIG. 1 . Data 15f includes the prediction model 1g, information on the type of the second structure 3A associated with a utility pole, information on the disaster 4f associated with an earthquake, and information on the damage factor 5f associated with vibrations. Data 15g includes the prediction model 1g, information on the type of the second structure 3G associated with a communication pipe, information on the disaster 4f associated with an earthquake, and information on the damage factor 5g associated with liquefaction. In other words, the prediction model 1f is linked to the same disaster 4a as the target disaster associated with an earthquake, and the same damage factor 5f as the target factor associated with vibrations. Furthermore, the prediction model 1g is linked to the same disaster 4a as the target disaster of the earthquake and the same damage factor 5g as the target factor of the liquefaction. In this case, in the processing of step S2, the control unit 16 calls the prediction model 1f and the prediction model 1g.

[0067] The processing of steps S3 and S4 according to the second modified example is the same as or similar to the processing of steps S3 and S4 according to the first modified example.

[0068] <Third Modification> The size of the shape data of the first structure 2 is smaller than the size of the mesh M constituting the prediction model 1 of the second structure 3. ｉ If the size is smaller than the size of the image, the control unit 16 may execute the processes of steps S3 to S6 according to the third modified example as follows.

[0069] In the process of step S3 according to the third modification, the control unit 16 virtually places the second structure 3 on the first structure 2 by placing an object of the second structure 3 around the shape data of the first structure 2. In this case, the actual position of the first structure 2 is calculated based on the position of each mesh M of the prediction model 1 of the second structure 3. ｉ , the position may not be included in the range of positions associated with the

[0070] For example, in FIG. 4A , the first structure 2 is a utility pole. The second structure 3 is a road. In FIG. 4A , data is used to represent the damage situation when the first structure 2, which is a utility pole, is damaged by a landslide caused by heavy rain, and when the second structure 3, which is a road, is damaged by a landslide caused by heavy rain and the road is blocked. The size of the shape data 120 of the first structure 2, which is a utility pole as shown on the left side of FIG. 4A , is the mesh M of the prediction model 1 of the second structure 3. ｉ As shown on the right side of FIG. 4A , the control unit 16 arranges an object 121 of the second structure 3 around the shape data 120 of the first structure 2. The dashed line on the right side of FIG. 4A indicates the mesh M of the prediction model 1 of the second structure 3. ｉ Shows.

[0071] In the process of step S4 according to the third modification, the control unit 16 calculates the damage rate R ｉ Variable X of １ ~X Ｌ The prediction model 1 is applied to the second structure 3 by substituting values ​​based on the environmental settings of the first structure 2, etc.

[0072] For example, Fig. 4B shows a configuration in which prediction model 1 is applied to a second structure 3 virtually placed on a first structure 2. In Fig. 4B, objects 121 with a high damage rate are colored darker, and objects 121 with a low damage rate are colored lighter.

[0073] In the process of step S6 according to the third modification, the control unit 16 calculates the damage rate R ｉ may be obtained as the damage rate of the first structure 2.

[0074] For example, in FIG. 4C, the control unit 16 calculates the damage rate R ｉ is obtained as the damage rate of the first structure 2.

[0075] However, the processes of steps S3 to S6 according to the third modification are not limited to the above-described processes. As another example, when the actual position of the first structure 2 is calculated based on each mesh M of the prediction model 1 of the second structure 3, ｉ In this case, the control unit 16 may use the prediction model 1 of the second structure 3 that corresponds to the actual position of the first structure 2 without performing the process of step S3. The control unit 16 may perform the process of step S4 on the prediction model 1 of the second structure 3 that corresponds to the actual position of the first structure 2.

[0076] (Example of Generation of Prediction Model) A description will be given of an example of a method for generating the prediction model 1 of the second structure 3. The prediction model 1 of the second structure 3 may be generated by the prediction device 10 or by another device.

[0077] The prediction model 1 is generated using damage data obtained when the second structure 3 is actually damaged. Multiple pieces of damage data obtained when the second structure 3 is hit by different disasters are collectively referred to as a "disaster data group." The multiple pieces of damage data included in the damage data group are classified by damage cause 5.

[0078] For example, in FIG. 5, the damage data included in the damage data group 30 is classified into damage data 31 of damage cause 5A, damage data 32 of damage cause 5B, and damage data 33 of damage cause 5C.

[0079] Among the multiple pieces of damage data included in the damage data group, even if the damage data is caused by a single disaster, the damage data caused by multiple damage factors may be classified into each of the multiple damage factors. For example, in FIG. 5, the damage data included in the damage data group 30 (hereinafter referred to as "damage data 34") is data generated when the second structure 3 is affected by a single disaster, but is caused by both damage factors 5A and 5B. In this case, the damage data 34 is classified into both damage factors 5A and 5B.

[0080] A prediction model 1 is constructed from the classified damage data, data on the disaster suffered by the second structure 3, and explanatory variables set according to the set environment or the configuration of the second structure 3. The prediction model 1 is constructed by, for example, machine learning or a statistical method. For example, the prediction model 1 may be generated by the method described in Non-Patent Document 1.

[0081] As described above, in the prediction device 10 according to the present embodiment, the control unit 16 calls up from the memory unit 13 a prediction model 1 that is linked to a disaster 4 of the same type as the target disaster type received from the input unit 11 and to a damage factor 5 that is the same as the target factor. The control unit 16 acquires a prediction result of the damage situation of the first structure 2 based on the called prediction model 1. Here, if the types of the first structure 2 and the second structure 3 are different, even if the same damage factor occurs in the same disaster, the damage suffered by the first structure 2 and the damage suffered by the second structure 3 will be different. For example, assume that the first structure 2 is a road and the second structure 3 is a utility pole. In this case, if damage factors of fallen trees and road collapse occur in heavy rain, the damage suffered by the first structure 2 will be road closure, and the damage suffered by the second structure 3 will be a tilted utility pole. However, if the same damage factor occurs in the same disaster, the same damage factor that occurs in the same disaster will be a factor that causes damage to both the first structure 2 and the second structure 3. Therefore, even if the type of the first structure 2 and the type of the second structure 3 are different, the damage status of the first structure 2 can be predicted using a prediction model that is linked to the same type of disaster 4 as the type of target disaster and to the same damage factor 5 as the target factor. With this configuration, this embodiment can predict the damage status of structures for which sufficient damage data is not available. As a result, this embodiment makes it possible to predict the macro-level damage status. Furthermore, after predicting the damage status, it becomes possible to take preventive measures, etc.

[0082] [Supplementary Item 1] A prediction device comprising: a memory unit that stores at least one prediction model, the prediction model being linked to information on the type of disaster and information on damage factors; an input unit that accepts information on the type of target disaster and information on target factors; and a control unit, wherein the target disaster is a disaster that is predicted to occur to a first structure whose damage situation is to be predicted, the target factors are factors of damage that are predicted to occur when the first structure is hit by the target disaster, the prediction model being generated from data of a second structure different from the first structure, and the control unit calls up from the memory unit the prediction model that is linked to the disaster of the same type as the type of target disaster and to the same damage factors as the target factors, and obtains a prediction result for the damage situation of the first structure based on the called prediction model.

[0083] [Supplementary Item 2] The prediction device according to Supplementary Item 1, further comprising a display unit, wherein the control unit causes the display unit to display a prediction result of the damage situation of the first structure.

[0084] [Supplementary Item 3] The prediction model includes topographical data divided into a plurality of meshes and data on the damage rate of the second structure in each mesh of the topographical data, and the control unit virtually places the second structure of the called prediction model on the first structure, obtains the damage rate of the second structure in each mesh by applying the called prediction model to the second structure virtually placed on the first structure, and obtains the damage rate of the first structure based on the obtained damage rate of the second structure as a prediction result of the damage situation of the first structure. This is the prediction device described in Supplementary Item 1 or 2.

[0085] [Supplementary Item 4] The prediction device described in any one of Supplementary Items 1 to 3, wherein the control unit sets a calculation range for calculating the damage rate of the first structure in the shape data of the first structure, and calculates the damage rate of the first structure for each of the calculation ranges as a prediction result of the damage situation of the first structure.

[0086] The present disclosure is not limited to the above-described embodiments. For example, two or more blocks shown in the block diagram may be integrated, or one block may be divided. Two or more steps shown in the flowchart may be executed in parallel or in a different order, instead of being executed in chronological order as described, depending on the processing capabilities of the device executing each step, or as needed. Other modifications are possible within the scope of the present disclosure.

[0087] For example, the memory unit 13 may store shape data of the first structure 2 linked to information on the type of the first structure 2. In this case, in the processing of step S1, the control unit 16 may not accept the shape data of the first structure 2. Instead, before the processing of step S5, the control unit 16 may acquire, from the memory unit 13, the shape data of the first structure 2 linked to the information on the type of the first structure 2 accepted in the processing of step S1.

[0088] The prediction device of the present disclosure can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided via a network.

[0089] For example, an embodiment is possible in which a general-purpose computer functions as the prediction device 10 according to the above-described embodiment. Specifically, a program describing the processing content for realizing each function of the prediction device 10 according to the above-described embodiment is stored in the memory of the general-purpose computer, and the program is read and executed by a processor. Therefore, the present disclosure can also be realized as a program executable by a processor or a non-transitory computer-readable medium storing the program.

[0090] 1, 1a, 1b, 1c, 1d: Prediction model 2: First structure 3, 3A, 3B: Second structure 4, 4a, 4b: Disaster 5, 5a, 5b, 5c, 5A, 5B, 5C: Damage factor 10: Prediction device 11: Input unit 12: Display unit 21, 121: Object 13: Memory unit 14: Data group 15, 15a, 15b, 15c, 15d: Data 16: Control unit 20: Shape data 21, 121: Object 30: Damage data group 31, 32, 33, 34: Damage data 120M: Mesh

Claims

1. A prediction device comprising: a memory unit that stores at least one prediction model, the prediction model being linked to information on the type of disaster and information on damage factors; an input unit that accepts information on the type of target disaster and information on target factors; and a control unit, wherein the target disaster is a disaster that is predicted to occur to a first structure whose damage situation is to be predicted; the target factors are factors of damage that are predicted to occur when the first structure is hit by the target disaster; the prediction model is generated from data of a second structure different from the first structure; and the control unit calls from the memory unit the prediction model that is linked to the same type of disaster as the type of target disaster and to the same damage factors as the target factors, and obtains a prediction result for the damage situation of the first structure based on the called prediction model.

2. The prediction device according to claim 1, further comprising a display unit, wherein the control unit causes the display unit to display the predicted results of the damage situation of the first structure.

3. The prediction device described in claim 1 or 2, wherein the prediction model includes topographical data divided into a plurality of meshes and data on the damage rate of the second structure in each mesh of the topographical data, and the control unit virtually places the second structure of the called prediction model on the first structure, obtains the damage rate of the second structure in each mesh by applying the called prediction model to the second structure virtually placed on the first structure, and obtains the damage rate of the first structure based on the obtained damage rate of the second structure as a prediction result of the damage situation of the first structure.

4. The prediction device described in claim 3, wherein the control unit sets a calculation range for calculating the damage rate of the first structure in the shape data of the first structure, and calculates the damage rate of the first structure for each calculation range as a prediction result of the damage situation of the first structure.

Images