Building damage estimation method, building damage estimation system, and program

Abstract

To provide a method for estimating the degree of damage of a building, a system for estimating the degree of damage of the building, and a program, capable of improving accuracy.SOLUTION: First earthquake information at a plurality of first observation points and second earthquake information of fewer types than the first earthquake information at a plurality of second observation points including the first observation points and other points are acquired. At each of the first observation points, first result information is acquired as output information by a first output derivation function using the first earthquake information as input information and predetermined earthquake result information as output information, and second result information is acquired as output information by a second output derivation function using the second earthquake information as input information and the predetermined earthquake result information as output information. A correction function of the second result information is derived from the first result information and the second result information at the first observation position. At the second observation point, the second result information is corrected by the correction function to acquire second corrected result information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a method for estimating the damage level of a building, a system for estimating the damage level of a building, and a program. [Background technology]

[0002] Patent Document 1 discloses a building damage estimation system and method. This building damage estimation system is composed of a building damage estimation device, earthquake sensors installed on the building's foundation ground surface, and earthquake sensors installed at representative points within the building. The building damage estimation device includes a virtual building model derivation unit that derives a virtual building model that mathematically represents the earthquake-induced movement of the building based on known building information about the building, an earthquake response analysis unit that inputs ground acceleration measured by the earthquake sensors during an earthquake into the virtual building model and performs earthquake response analysis of the virtual building model, an earthquake and movement information calculation unit that uses the results of the earthquake response analysis to calculate earthquake information and movement information that indicates the movement of the building due to the earthquake, and a damage estimation unit that uses the earthquake information and movement information after the earthquake has ended to estimate the damage level of the building.

[0003] The building damage estimation system and method described in Patent Document 1 does not require the installation of acceleration sensors on every floor of a building, but only requires the installation of earthquake sensors on the foundation ground surface and at representative points within the building, making it possible to estimate the degree of damage to a building after an earthquake more cheaply than conventional methods.

[0004] However, the building damage estimation system and method described in Patent Document 1 requires the installation of earthquake sensors at least at representative points within the building, which, although cheaper than installing acceleration sensors on all floors of the building, still poses the problem of making it difficult to reduce costs.

[0005] Furthermore, the earthquake response analysis performed by the earthquake response analysis department is specifically a time history response analysis that takes into account the nonlinearity of building rigidity, which poses the problem of time-consuming analysis.

[0006] Therefore, a building damage estimation method described in Patent Document 2 was developed. The building damage estimation method described in Patent Document 2 does not require information on time-series waveforms of earthquake motion from multiple earthquake sensors installed in actual dwelling units. Instead, it uses as input information the feature quantities of earthquake motion (such as information on the maximum acceleration of earthquake motion and spectral values ​​in a response spectrum analysis of the time-series waveforms of earthquake motion) obtained from time-series waveform data recorded by seismometers installed around the building whose damage is to be estimated. This eliminates the need to install any or very few earthquake sensors in the building, thereby reducing costs. Furthermore, because the damage to a building is estimated without performing a time history response analysis that takes into account the nonlinearity of building rigidity, the time required to estimate the damage to a building is shortened. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-197013 [Patent Document 2] Japanese Patent Application Publication No. 2022-148889 Summary of the Invention [Problem to be solved by the invention]

[0008] Time-series waveform data of earthquake motion can be obtained, for example, from the "Strong Motion Observation Network (K-NET, KiK-net)" system operated by the National Research Institute for Earth Science and Disaster Resilience. While the "Strong Motion Observation Network (K-NET, KiK-net)" can acquire data from numerous observation points across the country, adjacent observation points are located tens of kilometers apart. Therefore, time-series waveform data for target points for damage estimation can only be acquired if there is an observation point nearby. Therefore, target points for damage estimation with a certain degree of accuracy are scattered. Therefore, it is possible to use a system such as "QUIET+ (registered trademark)" operated by the Structural Planning Institute, which has a shorter distance between adjacent observation points than the "Strong Motion Observation Network (K-NET, KiK-net)." In "QUIET+ (registered trademark)," observation points are arranged in a 250-meter mesh, allowing for planar and continuous coverage of target points for damage estimation. However, with "QUIET+ (registered trademark)," it is not possible to obtain time-series waveform data at observation points, and the only data that can be obtained is limited data such as measured seismic intensity, maximum acceleration, and maximum velocity. For this reason, when using "QUIET+ (registered trademark)," the target points can be captured planarly and continuously, making it possible to estimate the extent of damage more accurately than when using "strong motion observation networks (K-NET, KiK-net)." However, in terms of obtainable data, the data items are limited, so there is a risk that the accuracy will be lower than when using "strong motion observation networks (K-NET, KiK-net)."

[0009] The present disclosure has been invented in consideration of the above-mentioned conventional problems, and aims to provide a building damage estimation method, building damage estimation system, and program that obtain output information with improved accuracy at multiple observation points by utilizing highly accurate output information obtained from detailed input information at a small number of observation points to correct output information obtained from limited input information at a large number of observation points. [Means for solving the problem]

[0010] A building damage estimation method according to one embodiment of the present disclosure includes a first step, a second step, a third step, and a fourth step. The first step is a step of acquiring first earthquake information and second earthquake information. The first earthquake information consists of information on a plurality of first-type items related to seismic motion at a plurality of first observation points. The second earthquake information consists of information on a plurality of second-type items related to the seismic motion that are included in the plurality of first-type items but are fewer than the plurality of first-type items at a plurality of second observation points that include the plurality of first observation points and also include observation points other than the plurality of first observation points. The second step is a step of acquiring first result information and second result information at each of the plurality of first observation points. The first result information is obtained as output information by a first output derivation function that uses the first earthquake information as input information and that uses predetermined earthquake result information as output information. The second result information is obtained as output information by a second output derivation function that uses the second earthquake information as input information and that uses the predetermined earthquake result information as output information. The third step is a step of deriving a correction function. The correction function is a function that, from the first result information and the second result information at each of the plurality of first observation points, approaches the second result information to the correct answer, assuming the first result information to be a correct answer. The fourth step is a step of acquiring second corrected result information. The second corrected result information is information obtained by correcting the second result information derived by the second output derivation function at each of the plurality of second observation points using the correction function.

[0011] A building damage estimation system according to one embodiment of the present disclosure includes a first processing unit, a second processing unit, a third processing unit, and a fourth processing unit. The first processing unit acquires first earthquake information and second earthquake information. The first earthquake information comprises information on a plurality of first-type items related to seismic motion at a plurality of first observation points. The second earthquake information comprises information on a plurality of second-type items included in the plurality of first-type items related to the seismic motion but fewer than the plurality of first-type items at a plurality of second observation points, the second observation points including the plurality of first observation points and including observation points other than the plurality of first observation points. The second processing unit acquires first result information and second result information for each of the plurality of first observation points. The first result information is obtained as output information by a first output derivation function that uses the first earthquake information as input information and predetermined earthquake result information as output information. The second result information is obtained as output information by a second output derivation function that uses the second earthquake information as input information and the predetermined earthquake result information as output information. The third processing unit derives a correction function. The correction function is a function that, based on the first result information and the second result information at each of the plurality of first observation points, approaches the second result information to the correct answer, with the first result information being considered as the correct answer. The fourth processing unit acquires second corrected result information. The second corrected result information is information obtained by correcting, with the correction function, the second result information derived by the second output derivation function at each of the plurality of second observation points.

[0012] A program according to one aspect of the present disclosure causes one or more processors to execute the method for estimating the damage level of a building. [Effects of the Invention]

[0013] According to one aspect of the building damage estimation method, building damage estimation system, and program of the present disclosure, highly accurate output information obtained from detailed input information at a small number of observation points can be used to correct output information obtained from limited input information at a large number of observation points, thereby obtaining output information with improved accuracy at a large number of observation points. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a block diagram of a building damage degree estimation system and a damage degree learning system according to an embodiment of the present disclosure. [Figure 2] FIG. 2 is a configuration diagram of a neural network in a trained model in the embodiment. [Figure 3] FIG. 3 is a configuration diagram of an analytical model of a building in the embodiment. [Figure 4] Fig. 4A is a diagram showing a graph comparing the first result information (dck) and the second result information (dpk) in the above-mentioned building damage estimation method. Fig. 4B is a diagram showing a graph comparing the first result information (dck) and the second corrected result information (^dpk) in the above-mentioned building damage estimation method. [Figure 5] FIG. 5 is a histogram showing the distribution of spatial data Zk in the above-mentioned method for estimating the damage level of a building. [Figure 6] Fig. 6A is a diagram showing the relationship between the predicted sum of squares PSSi of the second correction result information ^dpk and the range ai, and Fig. 6B is a diagram showing the relationship between the predicted sum of squares PSSi of the spatial data ^Zk and the range ai. [Figure 7] Figure 7A shows the relationship between the distance h and the theoretical variogram, variogram cloud, and empirical variogram. Figure 7B shows the relationship between the distance h and the theoretical variogram, variance of Z, and empirical variogram. DETAILED DESCRIPTION OF THE INVENTION

[0015] (overview) The present disclosure relates to a building damage estimation method, a building damage estimation system, and a program. The building damage estimation method of the present disclosure aims to obtain output information with improved accuracy at multiple observation points by using highly accurate output information obtained from detailed input information at a small number of observation points to correct output information obtained from limited input information at multiple observation points, while assuming that the method is capable of quickly estimating the damage level of a building when an earthquake occurs and achieving rapid and low-cost estimation of the damage level of a building.

[0016] Human casualties and damage to buildings caused by earthquakes have become a bigger problem than before, and various efforts have been made in recent years to reduce these damages. Human casualties include injuries sustained by people falling or being struck by falling objects during and immediately after an earthquake, as well as injuries caused by damage to buildings. Furthermore, even if no human casualties occur during an earthquake, if damage to buildings occurs, there is a possibility that further damage or collapse of damaged buildings will result in human casualties. In either case, damage to buildings is not only damage to the building but also a cause of human casualties. Therefore, in order to minimize human casualties and damage to buildings, it is necessary to quickly grasp the degree of damage to the building (damage level).

[0017] For this reason, some companies, including home builders, have installed earthquake sensors in buildings to directly measure the vibrations of the building caused by earthquake motion, thereby assessing the degree of damage to the building. However, while this method makes it easy to accurately assess the degree of damage to a building, it is costly because it requires installing earthquake sensors in many dwelling units and establishing an environment for transmitting the measurement data from the earthquake sensors. Furthermore, the accuracy of the damage estimation is not good when it is based solely on the seismic intensity calculated from the seismic intensity distribution published by public institutions.

[0018] The building damage estimation method, building damage estimation system, and program disclosed herein do not require information on the time-series waveform of earthquake motion from multiple earthquake sensors installed in actual dwelling units as input information, and instead use as input information the feature quantities of earthquake motion (such as information on the maximum acceleration of earthquake motion and the spectral values ​​in a response spectrum analysis of the time-series waveform of earthquake motion) obtained from time-series waveform data recorded by seismometers installed around the building for which damage estimation is being performed. This eliminates the need to install any or very few earthquake sensors in the building, reducing costs, and also shortens the time required to estimate the damage to the building because the damage to the building can be estimated without performing a time-history response analysis that takes into account the nonlinearity of the building's rigidity.

[0019] The building damage degree estimation method disclosed herein comprises a first step, a second step, a third step, and a fourth step. The first step is a step of acquiring first earthquake information and second earthquake information as input information. The second step is a step of calculating first result information and second result information at each of a plurality of first observation points. The third step is a step of deriving a correction function. The fourth step is a step of calculating second corrected result information.

[0020] In this building damage degree estimation method, at a small number of first observation points, highly accurate first result information obtained by inputting detailed first earthquake information into a first output derivation function and less accurate second result information obtained by inputting limited second earthquake information into a second output derivation function are used to derive a correction function that brings the second result information closer to the first result information. Then, at a large number of second observation points, the less accurate second result information is corrected using the correction function, thereby obtaining second corrected result information with improved accuracy.

[0021] (detail) A building damage estimation method, a building damage estimation system, and a program according to an embodiment of the present disclosure will be described in detail below with reference to Figs. 1 to 3. The building damage estimation method of this embodiment can be executed by a building damage estimation system 1 shown in Fig. 1. The building damage estimation method includes a first step, a second step, a third step, and a fourth step. The damage estimation system 1 can be connected to a seismic motion waveform server 9 via a communication network 8.

[0022] (1) Damage Estimation System As described above, the damage degree estimation system 1 is a system for executing the building damage degree estimation method of this embodiment. The damage degree estimation system 1 includes a communication unit 11, a storage unit 12, and a processing unit 13. The damage degree estimation system 1 can be realized by one or more servers.

[0023] The communication unit 11 is a communication interface. The communication unit 11 is connectable to the communication network 8 and has a function of performing communication through the communication network 8. The communication unit 11 includes, for example, a transmitter and a receiver. The communication unit 11 complies with a predetermined communication protocol. The predetermined communication protocol can be selected from various well-known wired and wireless communication standards.

[0024] The storage unit 12 is used to store information used by the processing unit 13. The storage unit 12 includes one or more storage devices. Examples of the storage devices include a solid state drive (SSD), a hard disk drive (HDD), a random access memory (RAM), and an electrically erasable programmable read-only memory (EEPROM).

[0025] The processing unit 13 is a control circuit that controls the operation of the damage extent estimation system 1. The processing unit 13 can be realized, for example, by a computer system including one or more processors (microprocessors) and one or more memories. In other words, the one or more processors execute one or more (computer) programs (applications) stored in one or more memories to function as the processing unit 13. Here, the programs are pre-recorded in the memory of the processing unit 13, but they may also be provided via a telecommunications line such as the Internet or by being recorded on a non-transitory recording medium such as a memory card.

[0026] The processing unit 13 includes a pre-processing unit 130, a first processing unit 131, a second processing unit 132, a third processing unit 133, a fourth processing unit 134, and a presentation unit 135. The pre-processing unit 130, the first processing unit 131, the second processing unit 132, the third processing unit 133, the fourth processing unit 134, and the presentation unit 135 do not represent actual configurations, but represent functions realized by the processing unit 13.

[0027] (1-1) Preprocessing section The pre-processing unit 130 executes a step (input information preparation step) of preparing input information required in the damage estimation method executed by the damage estimation system 1. The input information consists of first earthquake information, which is input information for a first output derivation function described below, and second earthquake information, which is input information for a second output derivation function described below. In the input information preparation step, the pre-processing unit 130 stores the first earthquake information and the second earthquake information in the memory unit 12. The input information preparation step is not included in the steps of the building damage estimation method of this embodiment. Note that the input information preparation step is not included in the steps of the building damage estimation method in a narrow sense, but may be included in the building damage estimation method in a broad sense.

[0028] (1-1-1) First Earthquake Information The first output derivation function takes detailed first earthquake information as input information. The first earthquake information consists of information on a plurality of first-type items related to seismic motion at a plurality (small number) of first observation points. Specifically, the first earthquake information includes information on the maximum acceleration, maximum velocity, measured seismic intensity, duration of a predetermined acceleration, and spectral values ​​in a response spectrum analysis of the time-series waveform of the seismic motion.

[0029] In order to prepare the first earthquake information, the pre-processing unit 130 executes a process to obtain the first earthquake information based on time series data of displacement, time series data of velocity, or time series data of acceleration on the ground surface at the time of the occurrence of earthquake motion (collectively referred to as the time series waveform of earthquake motion).

[0030] The time-series waveforms of earthquake motion can be obtained from an earthquake motion waveform server 9 via a communication network 8. An example of the earthquake motion waveform server 9 is an internet server operated by the National Research Institute for Earth Science and Disaster Resilience, and a system called the "Strong Motion Observation Network (K-NET, KiK-net)" is available from this server as time-series waveforms of earthquake motion. The "Strong Motion Observation Network (K-NET, KiK-net)" can obtain data from approximately 1,700 observation points nationwide as primary observation points.

[0031] The maximum acceleration of earthquake motion is known as the Peak Ground Acceleration (PGA), which is the measured maximum acceleration at the ground surface when earthquake motion occurs. PGA can be calculated from the time-series waveform of earthquake motion (time-series acceleration data).

[0032] The maximum velocity of seismic motion is known as the Peak Ground Velocity (PGV), which is the measured maximum velocity at the ground surface when the seismic motion occurs. PGV can be calculated from the time-series waveform of the seismic motion (time-series velocity data).

[0033] The predetermined acceleration duration is the time during which the acceleration of the earthquake motion exceeds a predetermined acceleration obtained by multiplying PGA by a positive value less than 1. In this embodiment, the positive value by which PGA is multiplied is 0.8. The predetermined acceleration duration can be determined from the time-series waveform of the earthquake motion.

[0034] The spectral values ​​are obtained by response spectrum analysis of the time-series waveform of the acceleration of earthquake motion (in this embodiment, acceleration response spectrum analysis as shown below). Specifically, the target earthquake motion waveform is input into a linear one-mass system shear model with a constant damping constant (5% in this embodiment) and a time history response analysis is performed to obtain the maximum response acceleration, and by repeating this analysis while changing the natural period of the one-mass system shear model, an acceleration response spectrum that shows the relationship between the natural period and the maximum response acceleration is obtained.

[0035] The maximum acceleration and maximum velocity in the "Strong Motion Observation Network (K-NET, KiK-net)" are obtained from time-series waveform data in each of the three axial directions (east-west, north-south, and up-down) that the installed sensors can measure. From this, it is possible to select one horizontal direction necessary for estimating the extent of damage, or to vector-add data in two horizontal directions to extract the desired horizontal component.

[0036] (1-1-2) Second Earthquake Information The second output derivation function takes limited second earthquake information as input information. The second earthquake information consists of information on a plurality of second-type items related to seismic motion at a plurality (large number) of second observation points. The plurality of second-type items constituting the second earthquake information are included in the plurality of first-type items, and are fewer than the plurality of first-type items. Specifically, the second earthquake information includes information on the maximum acceleration, maximum velocity, and measured seismic intensity of the seismic motion. The second earthquake information does not include information on the specified acceleration duration and the spectral values ​​in the response spectrum analysis of the time-series waveform of the seismic motion.

[0037] To prepare the second earthquake information, the preprocessing unit 130 obtains the second earthquake information from the seismic waveform server 9 via the communication network 8. The seismic waveform server 9 may be, for example, an internet server operated by Structural Planning Institute, Inc. This server provides information on the maximum acceleration, maximum velocity, and measured seismic intensity of earthquake motion using a seismic motion map estimation system called "QUIET+ (registered trademark)." "QUIET+ (registered trademark)" uses earthquake records made public after an earthquake to estimate the measured seismic intensity equivalent value, maximum ground acceleration, and maximum ground velocity across Japan at a 250-m mesh resolution, and the results can be viewed on a web browser. In other words, by using "QUIET+ (registered trademark)," it is possible to obtain the maximum acceleration, maximum velocity, and measured seismic intensity at grid points spaced at 250-m intervals as second observation points.

[0038] The maximum acceleration and maximum velocity in QUIET+ (registered trademark) are calculated as a composite value in three directions. In other words, unlike the strong motion observation network (K-NET, KiK-net), it is not possible to obtain the desired horizontal component individually.

[0039] Although there may not be a second observation point that perfectly matches one first observation point, in "QUIET+ (registered trademark)" the estimation points for information on seismic motion are covered in a 250m mesh, so there is no problem in treating, for example, information on seismic motion at the estimation point closest to one first observation point as information on seismic motion at one first observation point. The multiple second observation points include multiple first observation points and also include multiple observation points other than the multiple first observation points.

[0040] (1-2) First processing section The first processing unit 131 executes the first step. The first step is a step of acquiring first earthquake information and second earthquake information. Here, "acquire" refers to taking data into a working area of ​​the memory of the processor of the processing unit 13 in order for the program to execute the building damage degree estimation method, and does not substantially include storing the data in an SSD or HDD.

[0041] The first processing unit 131 acquires from the storage unit 12 data of first earthquake information to be used as input information by the first output derivation function in the next second step, and places the data in a working area of ​​the memory of the processor. Also, the first processing unit 131 acquires from the storage unit 12 data of second earthquake information to be used as input information by the second output derivation function in the next second step, and places the data in a working area of ​​the memory of the processor.

[0042] (1-3) Second processing section The second processing unit 132 executes a second step. The second step is a step of calculating first result information and second result information at each of a plurality of first observation points. The first result information is predetermined earthquake result information obtained as output information by the first output derivation function. Specifically, the predetermined earthquake result information is information on the maximum deformation amount of the building.

[0043] The second result information is earthquake result information obtained as output information by the second output derivation function and has the same types of items as the first result information. The first result information and the second result information have the same target (item), the maximum deformation of the building, and should essentially produce the same result (value). However, the values ​​differ due to differences in the calculation accuracy caused by differences in the first and second output derivation functions used in the calculation and differences in the calculation methods for maximum acceleration and maximum velocity in the first and second earthquake information. The first output derivation function includes more items in the input information (first earthquake information) than the second output derivation function (second earthquake information), so the accuracy of the first result information is higher than the accuracy of the second result information. The first output derivation function and the second output derivation function are derived by machine learning, as will be described in detail later.

[0044] (1-4) Third processing section The third processing unit 133 executes a third step. The third step is a step of deriving a correction function. The correction function is a function that, based on the first result information and the second result information at each of the multiple first observation points, determines the first result information to be the correct answer and brings the second result information closer to the correct answer.

[0045] In this embodiment, first, in the third step, a plurality of first correction coefficients are calculated as correction functions at each of a plurality of first observation points. Next, a plurality of second correction coefficients are calculated as correction functions at each of a plurality of second observation points using the plurality of first correction coefficients at each of the plurality of first observation points by a Kriging technique. The derivation of the correction functions will be described in detail later.

[0046] (1-5) Fourth processing section The fourth processing unit 134 executes a fourth step. The fourth step is a step of calculating second correction result information. The second correction result information is information obtained by correcting, with a correction function, the second result information derived with the second output derivation function at each of the multiple second observation points. Specifically, the second correction result information is information on the maximum deformation amount of the same building as the second result information.

[0047] (1-6) Estimating the damage to buildings The damage level of the building is estimated based on the first result information at the first observation point obtained in the second step and the second corrected result information at the second observation point obtained in the fourth step.

[0048] (1-7) Presentation section The presentation unit 135 presents output information estimated by the building damage degree estimation method. In this embodiment, the output information estimated by the building damage degree estimation method is sent to the terminal device 7 via the communication network 8, so that the output information can be presented on the input / output unit 71 of the terminal device 7.

[0049] (2-1) Communication Network The communication network 8 may include the Internet. The communication network 8 may be configured not only of a network conforming to a single communication protocol, but also of multiple networks conforming to different communication protocols. The communication protocol may be selected from various well-known wired and wireless communication standards. The communication network 8 may include data communication devices such as repeater hubs, switching hubs, bridges, gateways, and routers.

[0050] (2-2) Terminal device The terminal device 7 can be used to display information from the building damage degree estimation system 1 and, for example, to transmit safety information to the damage degree estimation system 1. The terminal device 7 includes an input / output unit 71, a communication unit 72, and a processing unit 73. The terminal device 7 can be realized by a desktop computer, a laptop computer, or a mobile terminal (such as a smartphone, a tablet terminal, or a wearable terminal).

[0051] The input / output unit 71 includes an input device for operating the terminal device 7. The input device may include, for example, a keyboard, a mouse, a touchpad, etc. The input / output unit 71 also includes an image display device for displaying information. The image display device may include a thin display device such as a liquid crystal display or an organic EL display.

[0052] The communication unit 72 is a communication interface and is similar to the communication unit 11 described above, so a detailed description thereof will be omitted.

[0053] The processing unit 73 can be realized by, for example, a computer system including one or more processors (microprocessors) and one or more memories.

[0054] (3) Derivation of the first and second output derivative functions The first output derivation function and the second output derivation function are derived in advance by appropriate means before executing the building damage degree estimation method. In this embodiment, the first output derivation function and the second output derivation function are derived by machine learning using the damage degree learning system 2.

[0055] (3-1) Disaster Level Learning System The damage level learning system 2 includes a communication unit 21, a storage unit 22, and a processing unit 23. The damage level learning system 2 can be realized by one or more servers.

[0056] The communication unit 21 is a communication interface and is similar to the above-described communication unit 11, so a detailed description thereof will be omitted. The memory unit 22 is used to store information used by the processing unit 23 and is similar to the above-described memory unit 12, so a detailed description thereof will be omitted. The processing unit 23 is a control circuit that controls the operation of the damage level learning system 2, and has a hardware configuration similar to the above-described memory unit 12.

[0057] The processing unit 23 includes a preprocessing unit 231, an acquisition unit 232, a learning unit 233, and a presentation unit 234. The preprocessing unit 231, the acquisition unit 232, the learning unit 233, and the presentation unit 234 do not represent actual configurations, but represent functions realized by the processing unit 23.

[0058] (3-2) Preprocessing section The preprocessing unit 231 prepares input information (first earthquake information and second earthquake information) that serves as training data required for the damage level learning method executed by the damage level learning system 2. To prepare training data for the input information (first earthquake information) of the first output derivation function, the preprocessing unit 231 first acquires the time-series waveform of seismic motion. The time-series waveform of seismic motion that forms the basis of the training data can be acquired from the "Strong Motion Observation Network (K-NET, KiK-net)" system of the seismic motion waveform server 9 via the communication network 8, as in the damage level estimation system 1.

[0059] The preprocessing unit 231 obtains teacher data of the first earthquake information required for the first output derivation function from the time-series waveform of the earthquake motion acquired from the earthquake motion waveform server 9. The teacher data of the first earthquake information includes information on the PGA, PGV, measured seismic intensity, predetermined acceleration duration, and spectrum values ​​in response spectrum analysis, which are the same as the first earthquake information in the damage degree estimation system 1.

[0060] The preprocessing unit 231 also acquires teacher data for the input information (second earthquake information) of the second output derivation function from the system "QUIET+ (registered trademark)" of the seismic waveform server 9. The teacher data for the second earthquake information includes information on PGA, PGV, and instrumental seismic intensity similar to the second earthquake information in the damage degree estimation system 1.

[0061] The teacher data of the first earthquake information and the teacher data of the second earthquake information obtained by the preprocessing unit 231 are stored in the storage unit 22.

[0062] (3-3) Acquisition part In learning the first output derivation function, the acquisition unit 232 acquires teacher data of the first earthquake information from the storage unit 12 and places it in a working area of ​​the memory of the processor. In learning the second output derivation function, the acquisition unit 232 acquires teacher data of the second earthquake information from the storage unit 12 and places it in a working area of ​​the memory of the processor.

[0063] (3-4) Learning Department The learning unit 233 constructs a trained model by machine learning, which is a trained model that uses input information as input data and output information on the damage level of buildings due to earthquake motion as training data. The trained model uses the neural network 4 shown in Figure 2.

[0064] In this embodiment, in learning the first output derivation function, as input data, data on the measured seismic intensity is input to node 41000 of the input layer 41, data on PGA is input to node 41001, data on PGV is input to node 41002, and data on a predetermined acceleration duration (positive value 0.8) is input to node 41003. Furthermore, acceleration response spectral values ​​of the seismic motion are input to nodes 41004 to 41103 of the input layer 41. The spectral values ​​are: the maximum value in a periodic band of 0.0 (sec) or more and less than 0.1 (sec) to node 41004; the maximum value in a periodic band of 0.1 (sec) or more and less than 0.2 (sec) to node 41005; and the maximum value in a periodic band of 9.9 (sec) or more and less than 10.0 (sec) to node 41102. The input layer 41 has 104 nodes, from node 41000 to node 41103.

[0065] Neural network 4 has two intermediate layers, intermediate layer 42 and intermediate layer 43. Intermediate layer 42 has 64 nodes, node 4200 to node 4263. Intermediate layer 43 has 64 nodes, node 4300 to node 4363. Output layer 44 has one node 440. Output information output from node 440 is information on the maximum deformation amount of the building. The maximum deformation amount of the building used as training data is data obtained by performing a time history response analysis on the time series waveform of earthquake motion acquired by preprocessing unit 231 for the analytical model of the building. The analytical model of the building will be described later.

[0066] Furthermore, in learning the second output derivation function, although not shown in the drawing, as input data, measured seismic intensity data is input to node 41000 of input layer 41, PGA data is input to node 41001, and PGV data is input to node 41002. In learning the second output derivation function, nodes 41003 to 41103 used in learning the first output derivation function are not used.

[0067] The intermediate layer 42 has 64 nodes, ie, node 4200 to node 4263, as in the case of learning the first output derivation function, the intermediate layer 43 has 64 nodes, ie, node 4300 to node 4363, as in the case of learning the first output derivation function, and the output layer 44 has one node 440. The output information output from node 440 is information on the maximum deformation amount of the building.

[0068] (3-5) Presentation section 1, the presentation unit 234 presents various output results from the processing unit 23. In this embodiment, various output results from the processing unit 23 are sent to the terminal device 7 via the communication network 8, so that the output information can be presented on the input / output unit 71 of the terminal device 7.

[0069] (3-6) Building analysis model 3, in this embodiment, the analytical model of the building 3 is an n-mass system shear model with nonlinear shear stiffness and linear viscous damping. The only disturbance input to the building 3 is the vibration of the ground 30, and there is no influence of the rocking spring of the ground 30.

[0070] As shown in Figure 1, the mass of the i (i = 1 to n)th floor of building 3 is m i The stiffness of the i-th layer is expressed as k i , the layer attenuation coefficient of the i layer is c i , the absolute acceleration of the i-th layer at time t is u″ i (t), the story shear force Q at time t in the i-th story is i (t) is expressed by the following equation (1) based on the equation of motion. i (t) is the absolute velocity at time t, and u i (t) is the absolute displacement at time t.

[0071]

number

[0072] Inter-story displacement D of layer i at time t i (t)=u i (t)-u i -1(t), interlayer velocity D i ′(t)=u i ′(t)-u i -1′(t), interlayer acceleration D i ″(t)=u i ″(t)-u i -1"(t), equation (1) can be expressed as the following equation (2).

[0073]

number

[0074] The absolute acceleration u1"(t), absolute velocity u1'(t), and absolute displacement u1(t) of the first (i.e., i=1) story are obtained from the time-series waveform of the earthquake motion. Then, by performing a time history response analysis of the analytical model of Building 3, all the absolute accelerations ui ″(t), absolute velocity u i ′(t) and absolute displacement u i (t), all story displacements D i (t), interlayer velocity D i '(t) and inter-story acceleration D i ″, the maximum deformation of each layer is determined.

[0075] In this analytical model, the displacement, velocity, and acceleration are in one horizontal direction, and are independent in each direction, for example, the north-south direction and the east-west direction.

[0076] (3-7) Learning about the damage to buildings In the damage level learning system 2, first, a preprocessing step is executed. In the preprocessing step, the preprocessing unit 231 executes processing to obtain input information. Specifically, in the preprocessing step, data on measured seismic intensity is acquired. Furthermore, in the preprocessing step, time-series waveforms of earthquake motion are acquired, and data on PGA, PGV, and data on a predetermined acceleration duration are extracted. Furthermore, in the preprocessing step, the time-series waveforms of earthquake motion are subjected to acceleration response spectrum analysis, and spectral values ​​in each periodic band are acquired.

[0077] In this embodiment, the time series waveforms of 2,210 earthquakes that have occurred in Japan in recent years are obtained from the Strong Earthquake Observation Network (K-NET, KiK-net), a system operated by the National Research Institute for Earth Science and Disaster Resilience, and the damage level is learned.

[0078] Next, the acquisition step is executed. In the acquisition step, the acquisition unit 232 acquires input information.

[0079] Next, a learning step is executed. In the learning step, a trained model is constructed by the learning unit 233. In this embodiment, of the data based on the time-series waveforms of 2,210 earthquake motions, data based on the time-series waveforms of 1,657 earthquake motions, which corresponds to 75%, is used as training data, and data based on the time-series waveforms of 553 earthquake motions, which corresponds to 25%, is used as test data to confirm generalization ability.

[0080] Next, the presentation step is executed. In the presentation step, the presentation unit 234 presents the weights W1 to W3 of the neural network 4.

[0081] (4) Derivation of the correction function As described in (1-4) above, in the building damage degree estimation method, a correction function is derived in the third step. The correction function is intended to obtain second corrected result information closer to the correct answer by correcting the second result information with lower accuracy, assuming the first result information with higher accuracy as the correct answer. In this embodiment, the correction function is derived using the so-called Kriging method, which will be explained below.

[0082] (4-1) Covariance function and semivariogram (4-1-1) Acquisition of first result information at the first observation point First earthquake information is acquired at a first observation point (the kth point, k = 1, 2, ..., n) consisting of n points. Next, the first earthquake information is input to a first output derivation function to calculate first result information. The first result information at the kth point (k = 1, 2, ..., n) is the maximum deformation amount (maximum deformation amount) dc k is.

[0083] (4-1-2) Acquisition of second result information at the first observation point The second earthquake information is acquired at the first observation point (the kth point, k = 1, 2, ..., n). Next, the second earthquake information is input to the second output derivation function to calculate the second result information. The second result information at the kth point (k = 1, 2, ..., n) is the maximum deformation amount (maximum deformation amount) dp k is.

[0084] (4-1-3) Spatial data representing errors Spatial data Z representing the error of the second result information at the kth point (k=1, 2, ..., n) relative to the first result information k of, Z k =log 10 (dp k / dc k )...Equation (3) Defined as:

[0085] (4-1-4) Mean (expected value) and variance of spatial data Spatial Data Z k (k=1, 2, …, n) mean m and variance σ 2 Ask for.

[0086] (4-1-5) Derivation of covariance function and semivariogram Spatial Data Z k is regarded as the realization of the random variable Z(u) that depends on the position u of the observation point (u is a position vector), and its probabilistic characteristics are assumed to be quadratically stationary. In other words, it is assumed that there is a common mean m in the target area, and that the covariance of the random variables Z(u) and Z(u+h), which are separated by a distance h (h is a distance vector), depends only on the distance h (i.e., it does not depend on the position u). The covariance function C(h) and semivariogram γ(h) required for kriging are defined by equations (4) and (5). C(h) = C(0) - γ(h), C(0) = σ 2 ...Equation (4) γ(h)=(1 / 2)·Var[Z(u+h)-Z(u)]=(1 / 2)·E[{Z(u+h)-Z(u)} 2 ]...Equation (5) Here, the function E[x] is a function that represents the mean (expected value) of the random variable x, and the function Var[x] is a function that represents the variance of the random variable x.

[0087] Furthermore, we assume that the covariance function C(h) and the semivariogram γ(h) are isotropic, and that C(h) and γ(h) depend only on the absolute value of the distance h, h = |h|.

[0088] A Gaussian model is applied as the theoretical variogram, which is defined by equation (6). Theoretical variogram γ(h)=b+(cb)[1-exp{-(h / a) 2}]...Equation (6) Here, theoretically, nugget b=0, sill c=σ 2 Therefore, these are fixed values, and only the range a becomes the variable θ.

[0089] (4-2) Estimation of the random variable Z(u) using Kriging By determining the range a, nugget b, and sill c in the theoretical variogram γ(h), it becomes possible to estimate the random variable Z(u) using simple kriging.

[0090] (4-3) Setting the optimal range a (4-3-1) Predicted sum of squares (PSS) i A variable range (for example, 1≦a≦200 km, a is in 1 km units) is set for the variable θ (in this embodiment, it is range a, hereinafter referred to as range a), and each range a i (variable θ i ) Predicted sum of squares PSS i (Prediction Sum of Squares) is defined by equation (7). PSS i =Σ n k=1 (dc k -^dp ik ) 2 , ^dp ik =dp k ÷(10 ^Zik )...Equation (7) where ^Z ik range a to range a i Assuming that, the remaining (n-1) sets of spatial data Z excluding the kth point k The spatial data Z at the kth point obtained from the theoretical variogram γ(h) k is a tentative estimate of ^dp ik range a to range a iThis is provisional second correction result information at the k-th point when it is assumed that:

[0091] (4-3-2) Predicted sum of squares (PSS) i The minimum value of By varying the range a within the range of variation, each predicted sum of squares PSS i Calculate the predicted sum of squares PSS i The range a when takes the minimum value i a opt Set as.

[0092] This gives the predicted sum of squares PSS i is the minimum value, that is, the remaining (n-1) sets of spatial data Z excluding the k-th point k The second correction result information ^dp at the kth point obtained from the theoretical variogram γ(h) of k and the first result information dc at the kth point k and spatial data Z that minimizes the error between k It becomes possible to estimate

[0093] (5) Calculation of the second correction result information As explained in (4) above, in the third step, the spatial data ^Z at the second observation point (the jth point (j = 1, 2, ..., N)) j In the fourth step, the spatial data ^Z at the jth point (j = 1, 2, ..., N) can be estimated. j Using this, the second correction result information ^dp at the jth point j The second correction result information ^dp at the jth point is calculated. j is the second result information dp at the jth point j It is calculated from equation (8). ^dp j =dp j ÷(10 ^Zj )...Equation (8)

[0094] (6) Verification results In inventing the building damage estimation method of the present disclosure, various comparison simulations have been carried out to verify the validity and accuracy, which will be explained based on Figures 4A to 7B. The data used for the verification is data from the earthquake that occurred on the Noto Peninsula on January 1, 2024 (hereinafter referred to as the Noto Peninsula Earthquake). The first result information (dc k ) was obtained from the National Research Institute for Earth Science and Disaster Resilience's "Strong Earthquake Observation Network (K-NET, KiK-net)" and acceleration time history data (north-south and east-west directions) were used to calculate the maximum deformation amount dc k The second result information (dp k ) is calculated by the second output derivation function using the instrumental seismic intensity, maximum acceleration (PGA), and maximum velocity (PGV) obtained from the acceleration time history data (north-south and east-west directions) obtained from the National Research Institute for Earth Science and Disaster Resilience's "Strong Motion Observation Network (K-NET, KiK-net)." k The peak velocity (PGV) was calculated by integrating the time history of the acceleration obtained over time.

[0095] (6-1) Effect of correction function The effect of the correction using the correction function (fourth step) in the damage estimation method for a building disclosed herein was verified. FIG. 4A shows the first result information (dc k , north-south direction) and second result information (dp k , north-south direction). Also, FIG. 4B shows the first result information (dc k , north-south direction) and second correction result information (^dp k , north-south direction).

[0096] In FIG. 4A and FIG. 4B, the second result information (dp k ) and the second correction result information (^dp k ) is the first result information (dc k ), i.e., the closer it is to the dashed line in the graph, the better the accuracy. kand dc k In the region within 20 mm, the two tend to be roughly the same, and dp k and dc k As dp increases, the two tend to diverge. k and dc k Although the two tend to diverge as the value increases, the degree of divergence is smaller than in FIG. 4A, and it can be seen that the effect of the correction is apparent.

[0097] (6-2) Spatial Data Z k Distribution of Spatial Data Z k The distribution of spatial data Z k The mean m and standard deviation σ of the spatial data Z k The horizontal axis shows the spatial data Z k The vertical axis indicates frequency (frequency), and the dashed line indicates a normal distribution having the same area (area of ​​the area between the horizontal axis and the vertical axis) as the total area of ​​the histogram.

[0098] From Figure 5, spatial data Z k is roughly normally distributed, and the spatial data Z k It can be seen that can be treated as a random variable that follows a normal distribution.

[0099] (6-3) Predicted sum of squares (PSS) i Subject to Predicted sum of squares (PSS) in the third step i As in the present embodiment, the provisional second correction result information ^dp ik Predicted sum of squares PSS when using i (^dp ik ) and range a i (variable θ i ) is shown in Figure 6A. The predicted sum of squares PSS in the third step is i As the target of spatial data Z k Predicted sum of squares PSS when using i (Z k ) and range a i (variable θ i) is shown in Figure 6B.

[0100] As shown in Figure 6A, the predicted sum of squares PSS i As in this embodiment, the provisional second correction result information ^dp ik If you use the predicted sum of squares PSS i (^dp ik ) is range a i The minimum value is reached when the value is 7, and the range a i Within the range of fluctuation, range a i It is estimated that there exists an optimum value for

[0101] In contrast, as shown in Figure 6B, the predicted sum of squares PSS i The spatial data Z k If you use the predicted sum of squares PSS i (Z k ) is range a i = 1 is the minimum value, and range a i ≧1. Therefore, the second correction result information ^dp ik Predicted sum of squares PSS i When the range a was adopted as the target of i The optimal value of Z cannot be used to obtain the same results. k is the predicted sum of squares PSS i It is estimated that the area is not suitable for the purpose of

[0102] (6-4) Comparison of theoretical and empirical variograms Figure 7A shows the relationship between the distance h and the theoretical variogram, variogram cloud, and empirical variogram. Figure 7B shows the relationship between the distance h and the theoretical variogram, variance of Z, and empirical variogram. The theoretical variogram is calculated using the predicted sum of squares PSS i The range a when takes the minimum value i a opt7A and 7B show the same relationship between the distance h and the theoretical variogram and the empirical variogram, and only the range of the distance h is different.

[0103] From Figures 7A and 7B (especially Figure 7B), the theoretical variogram and the empirical variogram are roughly the same in terms of the variance of Z (σ 2 ) (= sill c), and sill c = σ 2 It can be seen that the setting is reasonable.

[0104] Also, the predicted sum of squares PSS from the empirical variogram i a when takes the minimum value opt It is difficult to identify the predicted sum of squares PSS i a when takes the minimum value opt It can be seen that adopting a method for searching for the above is useful for the building damage degree estimation method of the present disclosure.

[0105] (7) Summary of this embodiment In this building damage estimation method, first result information with higher accuracy than the first output derivation function is obtained at a small number of first observation points where detailed first earthquake information can be obtained, and second result information with lower accuracy than the second output derivation function is obtained at a large number of second observation points where only limited second earthquake information can be obtained. Although only low-accuracy second result information is obtained at a large number of second observation points, by deriving a correction function from the first result information and second result information at the first observation points, second corrected result information with higher accuracy than the second result information can be obtained at a large number of second observation points, improving the accuracy of building damage estimation.

[0106] (8) Variations The embodiments of the present disclosure are not limited to the above-described embodiments. The above-described embodiments can be modified in various ways depending on the design, etc., as long as the object of the present disclosure can be achieved. Modifications of the above-described embodiments are listed below. The modifications described below can be applied in appropriate combinations.

[0107] (8-1) The damage degree estimation system 1 and the damage degree learning system 2 in the present disclosure include, for example, a computer system. The computer system is primarily composed of a processor and memory as hardware. The functions of the building damage degree estimation method and the building damage degree learning method in the present disclosure are realized by the processor executing a program stored in the memory of the computer system. The program may be pre-stored in the memory of the computer system, provided via a telecommunications line, or provided in a non-transitory recording medium readable by the computer system, such as a memory card, optical disk, or hard disk drive. The processor of the computer system is composed of one or more electronic circuits, including a semiconductor integrated circuit (IC) or a large-scale integrated circuit (LSI). The integrated circuits, such as ICs and LSIs, are referred to by different names depending on the degree of integration, and include integrated circuits called system LSIs, very large-scale integrations (VLSIs), and ultra-large-scale integrations (ULSIs). Furthermore, field-programmable gate arrays (FPGAs), which are programmed after the LSI is manufactured, or logic devices capable of reconfiguring the connections within the LSI or the circuit partitions within the LSI, can also be used as processors. The electronic circuits may be integrated into one chip or distributed across multiple chips. The chips may be integrated into one device or distributed across multiple devices. The computer system referred to here includes a microcontroller having one or more processors and one or more memories. Therefore, the microcontroller is also composed of one or more electronic circuits including a semiconductor integrated circuit or a large-scale integrated circuit.

[0108] (8-2) Furthermore, multiple functions of the damage degree estimation system 1 and the damage degree learning system 2 may be integrated into one housing. Furthermore, at least some of the functions of the damage degree estimation system 1 and the damage degree learning system 2, for example, some of the functions of the processing unit 23, may be realized by the so-called cloud (cloud computing) or the like.

[0109] (8-3) The time series waveforms of earthquake motion published by public institutions are used as the time series waveforms of earthquake motion, but it is also possible to obtain time series waveforms of earthquake motion by measuring them yourself.

[0110] (8-4) The time-series waveform of earthquake motion does not necessarily have to be acquired via the communication network 8. For example, the damage degree estimation system 1 may acquire the time-series waveform of earthquake motion directly from the earthquake motion waveform server 9 without going through the communication network 8. Furthermore, the damage degree estimation system 1 may directly include a seismometer.

[0111] (8-5) The positive value by which PGA is multiplied does not have to be 0.8, but can be set appropriately to 0.9, 0.7, 0.6, 0.5, etc.

[0112] (8-6) The output information from the first output derivation function and the second output derivation function is not limited to information on the maximum deformation amount of the building.

[0113] (8-7) As the machine learning algorithm, an algorithm other than one that uses a neural network, such as a support vector machine, which is a supervised learning algorithm, may be used.

[0114] (8-8) The analytical model of Building 3 is not limited to an n-mass shear model with nonlinear shear stiffness and linear viscous damping.

[0115] (8-9) It is preferable that the input information to the first output derivation function and the second output derivation function includes information on the instrumental seismic intensity. This makes it easier to improve the accuracy of estimating the damage level of buildings.

[0116] (8-10) It is preferable that the output information from the first output derivation function includes information on the maximum deformation amounts for multiple stories, which makes it easier to improve the accuracy of estimating the damage level of the building.

[0117] (8-11) It is preferable that the output information from the first output derivation function includes information on the maximum inter-story deformation, which makes it easier to improve the accuracy of estimating the damage level of the building.

[0118] (8-12) It is preferable that the output information from the first output derivation function includes information on the cumulative plastic deformation of the building, which makes it easier to improve the accuracy of estimating the damage level of the building.

[0119] (8-13) It is preferable that the output information from the first output derivation function includes information on the damage rank, which makes it easier to improve the accuracy of estimating the damage level of the building.

[0120] (8-14) In addition to acceleration response spectrum analysis, other response spectrum analyses include velocity response spectrum analysis and displacement response spectrum analysis. Any of these can be used as the response spectrum analysis, and there are no particular restrictions on the response spectrum analysis.

[0121] (8-15) In the input information preparation process, it is not excluded that the first earthquake information and the second earthquake information acquired by the preprocessing unit 130 are stored in the working area of ​​the memory of the processor of the processing unit 13 rather than in the memory unit 12.

[0122] (8-16) It is not an essential requirement for the building damage estimation method and damage estimation system 1 disclosed herein that the first earthquake information includes information on the maximum acceleration, maximum velocity, measured seismic intensity, specified acceleration duration, and spectral values ​​in response spectrum analysis of the time-series waveform of the earthquake motion.

[0123] (8-17) It is not an essential requirement for the building damage estimation method and damage estimation system 1 disclosed herein that the second earthquake information include information on the maximum acceleration, maximum velocity, and measured seismic intensity of the seismic motion. Also, it is not an essential requirement for the building damage estimation method and damage estimation system 1 disclosed herein that the second earthquake information does not include information on the specified acceleration duration and the spectral value in the response spectrum analysis of the time-series waveform of the seismic motion.

[0124] (8-18) The time length and pitch of the periodic band of the spectral values ​​in the response spectrum analysis of the time series waveform of earthquake motion are not limited to 0.1 (sec) and can be set appropriately.

[0125] (8-19) In the first step, it is not excluded that the first earthquake information and the second earthquake information acquired by the first processing unit 131 are stored in an SSD or HDD.

[0126] (8-20) In the above embodiment, simple kriging is applied as the kriging, but the kriging to be applied does not have to be simple kriging, and may be, for example, ordinary kriging or universal kriging, and is not particularly limited.

[0127] (8-21) In the above embodiment, a Gaussian model is applied as a model of the theoretical variogram, but the model of the theoretical variogram applied does not have to be a Gaussian model, and may be, for example, a spherical function model, an exponential model, a Hall effect model, etc., and is not particularly limited.

[0128] (8-22) In the third step, if there is a first observation point whose first result information is equal to or less than a predetermined threshold, the first observation point may be excluded from the calculation of the first correction coefficient, or a predetermined value may be assigned as the first result information for the first observation point. This will be explained below.

[0129] 1st result information dc k Spatial data Z in a small range kis greatly affected by the estimation accuracy, and the value may vary widely. k > Spatial data Z with threshold dlim (for example, threshold dlim = 10 mm) k Only the first result information dc k First result information dc that is equal to or less than the threshold value dlim k The extracted spatial data Z k The third and fourth steps are performed using only the excluded first result information dc k Estimate of ^Z ik Alternatively, the first result information dc k Predicted sum of squares PSS while excluding i Calculate.

[0130] In this case, the excluded first result information dc k ≦threshold value dlim Second correction result information ^dp at the kth point k is not based on Kriging, and the validity of the correction is uncertain. k The absolute value of the estimation error caused by this is small, so it has little effect on the judgment of the degree of damage to the building. i The effect on the calculation results can also be ignored.

[0131] (8-23) In the above embodiment, in the step of determining the minimum value of the predicted sum of squares PSSi (see (4-3-2) above), only the range a was used as a variable of the theoretical variogram, and nugget b and sill c were used as fixed values ​​of the theoretical variogram. Alternatively, the range a and nugget b may be used as variables of the theoretical variogram, and only sill c may be used as a fixed value of the theoretical variogram. Alternatively, the range a and sill c may be used as variables of the theoretical variogram, and only nugget b may be used as a fixed value of the theoretical variogram. Alternatively, all of the range a, nugget b, and sill c may be used as variables of the theoretical variogram.

[0132] (9) Mode As is apparent from the above-described embodiments and modifications, the present disclosure includes the following aspects.

[0133] A first aspect of the building damage estimation method includes a first step, a second step, a third step, and a fourth step. The first step is a step of acquiring first earthquake information and second earthquake information. The first earthquake information consists of information on a plurality of first-type items related to seismic motion at a plurality of first observation points. The second earthquake information consists of information on a plurality of second-type items included in the plurality of first-type items related to seismic motion but fewer than the plurality of first-type items at a plurality of second observation points, which includes the plurality of first observation points and also includes observation points other than the plurality of first observation points. The second step is a step of acquiring first result information and second result information at each of the plurality of first observation points. The first result information is obtained as output information by a first output derivation function that uses the first earthquake information as input information and predetermined earthquake result information as output information. The second result information is obtained as output information by a second output derivation function that uses the second earthquake information as input information and predetermined earthquake result information as output information. The third step is a step of deriving a correction function. The correction function is a function that, based on the first result information and the second result information at each of the multiple first observation points, approaches the second result information to the correct answer, assuming the first result information to be the correct answer. The fourth step is a step of acquiring second corrected result information. The second corrected result information is information obtained by correcting the second result information derived by the second output derivation function at each of the multiple second observation points using the correction function.

[0134] In this building damage estimation method, first result information with higher accuracy than the first output derivation function is obtained at a small number of first observation points where detailed first earthquake information can be obtained, and second result information with lower accuracy than the second output derivation function is obtained at a large number of second observation points where only limited second earthquake information can be obtained. Although only low-accuracy second result information is obtained at a large number of second observation points, by deriving a correction function from the first result information and second result information at the first observation points, second corrected result information with higher accuracy than the second result information can be obtained at a large number of second observation points, improving the accuracy of building damage estimation.

[0135] A second aspect is a building damage estimation method based on the first aspect. In the second aspect, input information for the first output derivation function and the second output derivation function includes information on the maximum acceleration, maximum velocity, and instrumental seismic intensity of the earthquake motion.

[0136] According to this aspect, it becomes easier to efficiently obtain highly accurate output information such as the maximum deformation amount of a building.

[0137] A third aspect is a method for estimating the damage level of a building based on the first or second aspect. In the third aspect, the output information in the first output derivation function and the second output derivation function includes information on the maximum deformation amount of the building.

[0138] According to this aspect, the damage level can be estimated using information on the maximum deformation amount of the building, which is the most important factor in the damage level of the building.

[0139] A fourth aspect is a damage level estimation method for a building based on any one of the first to third aspects. In the fourth aspect, the input information for the first output derivation function includes information on a predetermined acceleration duration, and the input information for the second output derivation function does not include information on the predetermined acceleration duration. The predetermined acceleration duration is the time during which the acceleration of the earthquake motion continues to exceed a predetermined acceleration obtained by multiplying the maximum acceleration of the earthquake motion by a positive value less than 1, within a predetermined duration of the earthquake motion.

[0140] According to this aspect, it becomes easier to efficiently obtain highly accurate output information such as the maximum deformation amount of a building.

[0141] A fifth aspect is a method for estimating the damage level of a building based on any one of the first to fourth aspects. In the fifth aspect, the input information in the first output derivation function includes information on spectral values ​​in a response spectrum analysis of a time-series waveform of seismic motion, and the input information in the second output derivation function does not include information on spectral values.

[0142] According to this aspect, it becomes easier to efficiently obtain highly accurate output information such as the maximum deformation amount of a building.

[0143] A sixth aspect is a building damage degree estimation method based on any one of aspects 1 to 5. In the sixth aspect, in the third step, a plurality of first correction coefficients are calculated as correction functions at each of a plurality of first observation points, and a plurality of second correction coefficients are calculated as correction functions at each of a plurality of second observation points from the plurality of first correction coefficients using a Kriging technique.

[0144] According to this aspect, by using the Kriging technique, it is possible to efficiently and accurately interpolate the correction function.

[0145] The seventh aspect is a damage degree estimation method for a building based on the sixth aspect. In the seventh aspect, the first observation point is the k-th point (k=1, 2, ..., n). The second result information dp k First result information DC k The spatial data Z can be considered as the realization of a random variable depending on the position of the observation point, which represents the error relative to k , Z k =log 10 (dp k / dc k ) is defined as

[0146] Spatial data Z at each point at a distance h kDefine a theoretical variogram γ(h) that represents the spatial correlation of the variogram γ(h), define a variable θ that includes at least the range a among the multiple parameters that make up the theoretical variogram γ(h), and define the variable θ as the variable θ i Assuming that, the remaining (n-1) sets of spatial data Z excluding the kth point k The spatial data Z at the kth point obtained from the theoretical variogram γ(h) k Let the tentative estimate of ik Let's say.

[0147] Variable θ i variable θ i Assuming that, the provisional second correction result information ^dp at the kth point ik ^dp ik =dp k ÷(10 ^Zik ) is defined as

[0148] Each variable θ i Predicted sum of squares PSS i ,PSS i =Σ n k=1 (dc k -^dp ik ) 2 Defined as:

[0149] Predicted sum of squares PSS in the range of variation of the variable θ i The estimated value ^Z when the variable θ is at its minimum value k Using the second correction result information ^dp k Ask for.

[0150] An eighth aspect is a method for estimating the damage level of a building based on any one of aspects 1 to 7. In the seventh aspect, if there is a first observation point in the third step where the first result information is equal to or less than a predetermined threshold, this first observation point is excluded from the calculation of the first correction coefficient, or a predetermined value is assigned as the first result information for this first observation point.

[0151] According to this aspect, the influence of variations in the correction function can be suppressed.

[0152] A building damage estimation system according to a ninth aspect includes a first processing unit 131, a second processing unit 132, a third processing unit 133, and a fourth processing unit 134. The first processing unit 131 acquires first earthquake information and second earthquake information. The first earthquake information consists of information on a plurality of first-type items related to seismic motion at a plurality of first observation points. The second earthquake information consists of information on a plurality of second-type items that are included in the plurality of first-type items related to seismic motion but are fewer than the plurality of first-type items at a plurality of second observation points, the second observation points including the plurality of first observation points and including observation points other than the plurality of first observation points. The second processing unit 132 acquires first result information and second result information for each of the plurality of first observation points. The first result information is obtained as output information by a first output derivation function that uses the first earthquake information as input information and predetermined earthquake result information as output information. The second result information is obtained as output information by a second output derivation function that uses the second earthquake information as input information and predetermined earthquake result information as output information. The third processing unit 133 derives a correction function. The correction function is a function that, from the first result information and the second result information at each of the multiple first observation points, approaches the second result information to the correct answer, assuming the first result information to be the correct answer. The fourth processing unit 134 acquires second corrected result information. The second corrected result information is information obtained by correcting, with the correction function, the second result information derived by the second output derivation function at each of the multiple second observation points.

[0153] In this building damage estimation system, first result information with higher accuracy than the first output derivation function is obtained at a small number of first observation points where detailed first earthquake information can be obtained, and second result information with lower accuracy than the second output derivation function is obtained at a large number of second observation points where only limited second earthquake information can be obtained. Although only low-accuracy second result information is obtained at a large number of second observation points, by deriving a correction function from the first result information and second result information at the first observation points, second corrected result information with higher accuracy than the second result information can be obtained at a large number of second observation points, improving the accuracy of building damage estimation.

[0154] A program according to a tenth aspect causes one or more processors to execute the method for estimating the damage level of a building according to any one of the first to eighth aspects.

[0155] In this program, at a small number of first observation points where detailed first earthquake information can be obtained, first result information with higher accuracy is obtained from the first output derivation function, and at a large number of second observation points where only limited second earthquake information can be obtained, second result information with lower accuracy is obtained from the second output derivation function. At a large number of second observation points, only low-accuracy second result information is obtained, but by deriving a correction function from the first result information and second result information at the first observation points, second corrected result information with higher accuracy than the second result information can be obtained at a large number of second observation points, improving the accuracy of building damage estimation. [Explanation of symbols]

[0156] 1. Damage Estimation System 11 Communications Department 12 Storage section 13 Processing section 130 Pre-processing section 131 First Processing Section 132 Second Processing Section 133 Third Processing Section 134 4th Processing Section 135 Presentation section 2. Disaster Damage Learning System 21 Communications Department 22 Memory section 23 Processing section 231 Pretreatment section 232 Acquisition Department 233 Learning Department 234 Presentation section 3. Building 30 Ground 4. Neural Networks 41 Input layer 42 Middle Class 43 Middle Class 44 Output layer 7 Terminal Equipment 71 Input / output section 72 Communications Department 73 Processing section 8. Communication Networks 9 Earthquake Waveform Server

Claims

1. a first step of acquiring first earthquake information consisting of information on a plurality of first-type items related to seismic motion at a plurality of first observation points, and second earthquake information consisting of information on a plurality of second-type items related to the seismic motion that are included in the plurality of first-type items related to the seismic motion and that are fewer than the plurality of first-type items, at a plurality of second observation points that include the plurality of first observation points and also include observation points other than the plurality of first observation points; At each of the plurality of first observation points, A first output derivation function is used to obtain first result information as the output information, the first earthquake information being input information and predetermined earthquake result information being output information. a second step of acquiring second result information as the output information by a second output derivation function that uses the second earthquake information as input information and the predetermined earthquake result information as output information; a third step of deriving a correction function that approximates the second result information to the correct answer, based on the first result information and the second result information at each of the plurality of first observation points, with the first result information being regarded as a correct answer; and a fourth step of correcting the second result information derived by the second output derivation function at each of the plurality of second observation points by the correction function to obtain second corrected result information. Methods for estimating the extent of damage to buildings.

2. the input information for the first output derivation function and the second output derivation function includes information on a maximum acceleration, a maximum velocity, and an instrumental seismic intensity of the seismic motion; The method for estimating the damage level of a building according to claim 1.

3. the output information in the first output derivation function and the second output derivation function includes information on a maximum deformation amount of the building; The method for estimating the damage level of a building according to claim 1.

4. the input information in the first output derivation function includes information on a predetermined acceleration duration, and the input information in the second output derivation function does not include information on the predetermined acceleration duration; The predetermined acceleration duration is a time during which the acceleration of the earthquake motion exceeds a predetermined acceleration obtained by multiplying the maximum acceleration of the earthquake motion by a positive value less than 1, within the predetermined duration of the earthquake motion. The method for estimating the damage level of a building according to claim 1.

5. the input information in the first output derivation function includes information on a spectral value in a response spectrum analysis of a time-series waveform of the seismic motion, and the input information in the second output derivation function does not include information on the spectral value. The method for estimating the damage level of a building according to claim 1.

6. In the third step, calculating a plurality of first correction coefficients as the correction function at each of the plurality of first observation points; calculating a plurality of second correction coefficients as the correction function at each of the plurality of second observation points from the plurality of first correction coefficients by a Kriging method; The method for estimating the damage level of a building according to claim 1.

7. The first observation point is the kth point (k=1, 2, ..., n), The second result information dp ｋ The first result information dc ｋ The spatial data Z can be considered as the realization of a random variable depending on the position of the observation point, which represents the error for ｋ of, Z ｋ =log １０ (dp ｋ / dc ｋ ) Defined in The spatial data Z at each point at a distance h ｋ Define the theoretical variogram γ(h) that represents the spatial correlation of Among the multiple parameters constituting the theoretical variogram γ(h), a variable θ including at least the range a is defined; The variable θ is changed to variable θ ｉ In this case, the remaining (n-1) sets of spatial data Z excluding the k point are ｋ The spatial data Z at the k point obtained from the theoretical variogram γ(h) ｋ The provisional estimate of ｉｋ year, The variable θ is ｉ When it is assumed that the second correction result information ^dp at the point k is ｉｋ of, ^dp ｉｋ =dp ｋ ÷(10 ^Ｚｉｋ ) Defined in Each variable θ ｉ Predicted sum of squares PSS ｉ of, PSS ｉ =Σ ｎ ｋ＝１ (dc ｋ -^dp ｉｋ ) ２ Defined in The predicted sum of squares PSS ｉ The estimated value ^Z when the variable θ is at its minimum value ｋ Using the second correction result information ^dp ｋ Ask for The method for estimating the damage level of a building according to claim 6.

8. In the third step, if there is the first observation point at which the first result information is equal to or less than a predetermined threshold, This first observation point is excluded from the calculation of the first correction coefficient, or A predetermined value is assigned as the first result information at this first observation point. The method for estimating the damage level of a building according to claim 1.

9. a first processing unit that acquires first earthquake information consisting of information on a plurality of first-type items related to seismic motion at a plurality of first observation points, and second earthquake information consisting of information on a plurality of second-type items included in the plurality of first-type items related to the seismic motion and fewer than the plurality of first-type items at a plurality of second observation points that include the plurality of first observation points and also include observation points other than the plurality of first observation points; At each of the plurality of first observation points, A first output derivation function is used to obtain first result information as the output information, the first earthquake information being input information and predetermined earthquake result information being output information. a second processing unit that acquires second result information as the output information using a second output derivation function that uses the second earthquake information as input information and the predetermined earthquake result information as output information; a third processing unit that derives a correction function from the first result information and the second result information at each of the plurality of first observation points, with the first result information being a correct answer, to bring the second result information closer to the correct answer; a fourth processing unit that corrects the second result information derived by the second output derivation function at each of the plurality of second observation points by the correction function to obtain second corrected result information, Building damage estimation system.

10. one or more processors, Executing the building damage degree estimation method according to any one of claims 1 to 8, program.

Images