Earthquake motion data processing model generation method, earthquake motion data processing method, earthquake motion data processing model generation device, and earthquake motion data processing device

Abstract

To provide an earthquake data processing model creation method which can implement proper processing content with respect to earthquake time history data with incorporation of knowledge of an expert.SOLUTION: An earthquake data processing model creation method creates an earthquake data processing model 15 for processing earthquake time history data by machine learning using a computer. The earthquake data processing model creation method includes: an acquisition step of acquiring multiple pieces of learning data 14 made up of event data 12 and determination data 13 from a database 10 which stores event data 12 based on earthquake time history data 11 and determination data 13 including a determination result from determination of processing content which is implemented by an expert with on the earthquake time history data 11 in association with each other; and a creation step of creating the earthquake data processing model 15 as a machine-learned model by learning a correlation of the event data 12 and the determination data 13 on the basis of the plurality of pieces of the learning data 14.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to a seismic motion data processing model generating method, a seismic motion data processing method, a seismic motion data processing model generating device, and a seismic motion data processing device. [Background technology]

[0002] The Japanese archipelago and its surrounding areas are among the most earthquake-prone areas in the world, and especially since the 1995 Hyogo-ken Nanbu earthquake, a rare earthquake observation network has been deployed across the Japanese archipelago. The observation records of earthquake motion waveforms observed by the earthquake observation network are accumulated daily as earthquake motion time history data. Because of its usefulness, the earthquake motion time history data observed and accumulated in this way is widely used in society not only for the analysis and interpretation of earthquake motion characteristics, but also for clarifying the nature of earthquake source fault rupture, estimating underground structures, predicting the behavior and structural design of buildings and structures, and even for earthquake disaster prevention (see, for example, Patent Documents 1 and 2). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2015-215221 A [Patent Document 2] JP 2010-144487 A Summary of the Invention [Problem to be solved by the invention]

[0004] Before using earthquake motion time history data, it is necessary to carry out appropriate processing, such as removing inappropriate or disruptive data depending on the purpose of use, selecting data that meets the conditions, removing noise contained in the earthquake motion time history data, etc. In such cases, the task of determining the processing required for the earthquake motion time history data is generally left to the judgment of experts who are experienced in handling earthquake motion time history data.

[0005] However, due to recent changes in the environment, such as the rapid increase in data volume due to the expansion of earthquake observation networks, the fragmentation and sophistication of specialized fields, and the decrease in the number of experts due to the declining birthrate, the time and effort that individual experts can directly work on the judgment work of earthquake motion time history data is limited. In addition, although the recent improvement in the information processing capabilities of computers has the potential to greatly contribute to improving this problem, the construction of a system that can judge the necessary processing content for earthquake motion time history data and automatically process earthquake motion time history data has not yet been realized.

[0006] The present invention has been made in consideration of the above circumstances, and aims to provide a seismic motion data processing model generation method, a seismic motion data processing method, a seismic motion data processing model generation device, and a seismic motion data processing device that incorporate the knowledge of experts and enable appropriate processing to be performed on seismic motion time history data. [Means for solving the problem]

[0007] The present invention solves the above-mentioned problems, and a method for generating a seismic motion data processing model according to one embodiment of the present invention includes the steps of: A method for generating a seismic motion data processing model by machine learning using a computer to process seismic motion time history data, comprising: an acquisition step of acquiring a plurality of pieces of learning data composed of the case data and the judgment data from a database that stores case data based on the seismic motion time history data and judgment data including a judgment result when an expert judges a processing content to be performed on the seismic motion time history data in association with each other; and a generation process of generating the seismic motion data processing model as a trained model of the machine learning by learning the correlation between the case data and the judgment data by the machine learning based on the multiple learning data acquired in the acquisition process.

[0008] Moreover, the earthquake motion data processing model generating device according to an embodiment of the present invention comprises: The computer includes a control unit that executes each step included in the earthquake motion data processing model generation method.

[0009] Further, a seismic motion data processing method according to an embodiment of the present invention includes the steps of: A seismic motion data processing method for processing the seismic motion time history data based on the seismic motion data processing model generated by the seismic motion data processing model generation method using a computer, comprising the steps of: A receiving step of receiving the seismic motion time history data to be processed; a determination step of determining the processing content to be performed on the seismic motion time history data of the processing target based on the determination data output from the seismic motion data processing model by inputting the case data based on the seismic motion time history data of the processing target received in the receiving step into the seismic motion data processing model; Includes.

[0010] Moreover, a seismic motion data processing device according to an embodiment of the present invention comprises: The computer includes a control unit that executes each step included in the earthquake motion data processing method. Effect of the Invention

[0011] According to the earthquake motion data processing model generation method and earthquake motion data processing model device of one embodiment of the present invention, the knowledge gained by experts when determining the processing content to be performed on earthquake motion time history data can be incorporated into the earthquake motion data processing model through machine learning, making it possible to generate a earthquake motion data processing model that can determine appropriate processing content for earthquake motion time history data on behalf of experts.

[0012] Furthermore, according to the earthquake motion data processing method and earthquake motion data processing device of one embodiment of the present invention, the earthquake motion data processing model that has been learned from expert knowledge is used to determine the processing content required for the earthquake motion time history data to be processed, so that appropriate processing content can be implemented for the earthquake motion time history data to be processed on behalf of experts. [Brief description of the drawings]

[0013] [Figure 1] 1 is a schematic configuration diagram showing an example of a seismic motion time history data processing system 1 according to an embodiment of the present invention. [Diagram 2] FIG. 1 is a block diagram showing an example of a seismic motion time history data processing system 1 according to an embodiment of the present invention. [Diagram 3] FIG. 1 is a diagram showing an example of earthquake motion time history data 11 that is determined to be appropriate data by an expert. [Figure 4] FIG. 1 is a diagram showing an example of earthquake motion time history data 11 that is determined to be inappropriate by an expert. [Diagram 5] FIG. 2 is a diagram showing an example of a Fourier amplitude spectrum created based on earthquake motion time history data 11. [Figure 6] FIG. 2 is a functional explanatory diagram showing an example of a seismic motion data processing model generating device 4 and a seismic motion data processing model generating method according to the embodiment of the present invention. [Figure 7] FIG. 2 is a data structure diagram showing an example of a database 10. [Figure 8] FIG. 1 is a functional explanatory diagram showing an example of a seismic motion data processing device 5 and a seismic motion data processing method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings.

[0015] Fig. 1 is a schematic diagram showing an example of a seismic motion time history data processing system 1 according to an embodiment of the present invention. Fig. 2 is a block diagram showing an example of a seismic motion time history data processing system 1 according to an embodiment of the present invention.

[0016] The earthquake motion time history data processing system 1 includes an observation data providing device 2 that collects earthquake motion time history data 11 in which earthquake motions caused by earthquakes are observed at a plurality of observation points and provides the earthquake motion time history data 11 to an outside party, an expert terminal device 3 used by an expert on earthquakes, a earthquake motion data processing model generating device 4 that generates a earthquake motion data processing model 15 for processing the earthquake motion time history data 11 by machine learning, a earthquake motion data processing device 5 that processes the earthquake motion time history data 11 based on the earthquake motion data processing model 15 generated by the earthquake motion data processing model generating device 4, and a network 6 that connects the devices together. The network 6 communicates various data and signals by wireless communication or wired communication, and any communication standard is used.

[0017] The observation data providing device 2 collects earthquake motion time history data 11 of three components, for example, in the north-south direction, east-west direction, and up-down direction, measured by seismometers (not shown) installed at multiple observation points when an earthquake occurs, and provides the collected earthquake motion time history data 11 to an external party. The earthquake motion time history data 11 is, for example, data provided by the strong earthquake observation network K-NET of the National Research and Development Agency National Research Institute for Earth Science and Disaster Resilience (hereinafter referred to as "NIED"), and is data observed at 138 observation points (see FIG. 1) installed in one metropolis and six prefectures in the Kanto region (Tokyo, Kanagawa, Chiba, Saitama, Ibaraki, Tochigi, and Gunma) in the Kanto region shown in FIG. 1 as an example. In this case, the earthquake motion time history data 11 may be data observed for two horizontal components of an orthogonal coordinate system along axes specific to the observation points, instead of the two horizontal components in the north-south direction and the east-west direction.

[0018] In addition, when an earthquake occurs, the observation data providing device 2 may provide the earthquake motion time history data 11 related to the earthquake in real time to the earthquake motion data processing model generation device 4 or the earthquake motion data processing device 5, or when a data request is received from the earthquake motion data processing model generation device 4 or the earthquake motion data processing device 5, it may provide the earthquake motion time history data 11 related to the request (which may be earthquake motion time history data 11 related to a plurality of earthquakes that have occurred in the past and that meet specified conditions) to the earthquake motion data processing model generation device 4 or the earthquake motion data processing device 5.

[0019] The expert terminal device 3 is composed of, for example, a general-purpose or dedicated computer. The expert terminal device 3 is used when an expert evaluates seismic motion indicators related to the source and scale of an earthquake based on a seismic motion data group consisting of a plurality of seismic motion time history data 11. By analyzing the seismic motion data group, for example, the amplitude characteristics, periodic characteristics, and time characteristics of the seismic motion are evaluated as seismic motion indicators. Therefore, the seismic motion time history data 11 can be widely used in society not only for the analysis and interpretation of seismic motion characteristics, but also for clarifying the seismic source fault rupture characteristics of earthquakes, estimating underground structures, predicting the behavior and structural design of buildings and structures, and even for earthquake disaster prevention.

[0020] The earthquake motion time history data 11 is obtained by observing earthquake motion waveforms and recording them as data, but depending on the purpose of using the earthquake motion time history data 11, some of the data may be inappropriate or may cause problems. Therefore, experts need to perform processing such as excluding inappropriate data and removing noise when the data contains noise. Therefore, the expert terminal device 3 is configured to be able to display the earthquake motion time history data 11 via, for example, an analysis program or a browser, and when an expert determines the processing content (data removal processing or noise removal processing) to be performed on the earthquake motion time history data 11, it accepts an input of the expert's determination result, generates determination data 13 including the determination result, and provides it to the earthquake motion data processing model generation device 4.

[0021] 1, for the sake of simplicity, one expert terminal device 3 is illustrated, but there may be multiple expert terminal devices 3. In that case, each of the multiple expert terminal devices 3 provides the seismic motion data processing model generation device 4 with judgment data 13 including the judgment results by each expert.

[0022] FIG. 3 is a diagram showing an example of earthquake motion time history data 11 that is judged by an expert to be appropriate data. The earthquake motion time history data 11 is obtained by observing parameters (acceleration, velocity, or displacement) of north-south components, east-west components, and up-down components in time series at a predetermined sampling period (e.g., 0.01 seconds). As shown in FIG. 3(a), the P-wave portion corresponding to the initial motion of the earthquake motion contains many high-frequency components, and the S-wave portion corresponding to the main motion of the earthquake motion tends to have a large amplitude. In addition, the dominant period and waveform characteristics of subsequent motions often change in a variety of ways. In the earthquake motion time history data 11 observed at an observation point located close to the epicenter, the amplitude of the initial motion may be equal to or greater than that of the main motion, as shown in FIG. 3(b). In a large-scale earthquake, the complex and long-lasting fault rupture process affects the characteristics of the earthquake motion, as shown in FIG. 3(c).

[0023] FIG. 4 is a diagram showing an example of earthquake motion time history data 11 that is determined by an expert to be inappropriate data. The earthquake motion time history data 11 shown in FIG. 4(a) contains high-frequency noise throughout the entire waveform, so it is determined that noise removal processing is required to remove frequency noise in the frequency domain. At that time, frequency conditions when removing frequency noise with a frequency filter may be determined. The earthquake motion time history data 11 shown in FIG. 4(b) contains another earthquake or another event after subsequent motion convergence, so it is determined that noise removal processing is required to remove time history noise in the time domain. At that time, time conditions when removing time history noise with a time history filter may be determined.

[0024] The earthquake motion time history data 11 shown in Fig. 4(c) is determined to require data exclusion processing because it has a low S / N ratio (signal-to-noise ratio) and is missing records at the beginning of the time history waveform. The earthquake motion time history data 11 shown in Fig. 4(d) is determined to require data exclusion processing because it is missing records at the beginning and end of the time history waveform. The earthquake motion time history data 11 shown in Fig. 4(e) is determined to require data exclusion processing because it is missing records at the beginning of the time history waveform and another earthquake or another event is mixed in after the subsequent motion convergence.

[0025] Fig. 5 is a diagram showing an example of a Fourier amplitude spectrum created based on the earthquake motion time history data 11. The Fourier amplitude spectrum shown in Fig. 5 is a spectrum display of the Fourier amplitude obtained by Fourier transforming the earthquake motion time history data 11, and is displayed on a logarithmic axis with the period (or frequency) on the horizontal axis and the amplitude on the vertical axis.

[0026] The Fourier amplitude spectrum shown in Fig. 5(a) has few noise components and is judged by an expert to be appropriate data. On the other hand, the Fourier amplitude spectra shown in Fig. 5(b) and (c) contain a lot of low-frequency noise and high-frequency noise, respectively, and are therefore judged to require noise removal processing to remove frequency noise in the frequency domain. In this case, the frequency conditions when removing the frequency noise with a frequency filter may be determined.

[0027] (Configuration of the seismic motion data processing model generation device 4 and the processes performed by each part) The seismic motion data processing model generation device 4 associates case data 12 based on seismic motion time history data 11 provided from the observation data providing device 2 with judgment data 13 provided from the expert terminal device 3, and registers them in a database 10. The seismic motion data processing model generation device 4 refers to the database 10 and executes any machine learning algorithm, such as a gradient boosting tree, a random forest, a neural network, a convolutional neural network (CNN), or a recurrent neural network (RNN), to generate a seismic motion data processing model 15 as a trained model of machine learning. The seismic motion data processing model generation device 4 provides the seismic motion data processing model 15 to an external device (in this embodiment, the seismic motion data processing device 5).

[0028] The seismic motion data processing model generation device 4 is composed of a general-purpose or dedicated computer, and as shown in FIG. 2, is equipped with a storage unit 40 composed of a HDD, memory, etc., a control unit 41 composed of a processor such as a CPU, GPU, etc., a communication unit 42 which is a communication interface with the network 6, an input unit 43 composed of a keyboard, mouse, etc., and a display unit 44 composed of a display, touch panel, etc.

[0029] The storage unit 40 stores a database 10 in which case data 12 and judgment data 13 based on seismic motion time history data 11 are registered and updated, a seismic motion data processing model 15 which is a trained model, and a seismic motion data processing model generation program 400 which realizes the seismic motion data processing model generation method by controlling the operation of the seismic motion data processing model generation device 4. The database 10 may be stored in an external storage device instead of the storage unit 40, and in that case, the seismic motion data processing model generation device 4 may communicate with the external storage device via a network 6 to access the database 10.

[0030] The control unit 41 executes the earthquake motion data processing model generation program 400 to function as a DB management unit 410, an acquisition unit 411, and a generation unit 412. The functions of each unit will be described in detail later.

[0031] Fig. 6 is a functional explanatory diagram showing an example of the earthquake motion data processing model generating device 4 and the earthquake motion data processing model generating method according to the embodiment of the present invention. Fig. 7 is a data configuration diagram showing an example of the database 10.

[0032] (Database management process by DB management unit 410 and database 10) The DB management unit 410 receives the case data 12 based on the earthquake motion time history data 11 provided from the observation data providing device 2 and the judgment data 13 provided from the expert terminal device 3, and registers them in the database 10. Specifically, the DB management unit 410 converts the earthquake motion time history data 11 into case data 12, which is a predetermined data format, and registers them in the database 10. The DB management unit 410 also registers the judgment data 13, which includes the judgment result when the expert judges the processing content to be performed on the earthquake motion time history data 11, in the database 10 in association with the case data 12 based on the earthquake motion time history data 11. The DB management unit 410 may receive the case data 12 instead of the earthquake motion time history data 11. When receiving the judgment data 13 from the expert terminal device 3, the DB management unit 410 may also receive the earthquake motion time history data 11 (or the case data 12) corresponding to the judgment data 13.

[0033] As shown in FIG. 7, the database 10 stores a plurality of pairs of case data 12 and judgment data 13, in a state in which the case data 12 and judgment data 13 are associated with each other, the case data 12 being based on earthquake motion time history data 11.

[0034] The case data 12 is digital data based on the earthquake motion time history data 11, and may be, for example, numerical data for each time in which acceleration, velocity, or displacement are arranged in chronological order, or may be plotted data in vector or raster format, or any combination thereof. Moreover, instead of the earthquake motion time history data 11, the case data 12 may be data in which a Fourier spectrum (for example, the Fourier amplitude spectrum shown in FIG. 5) is recorded, or data in which a response spectrum is recorded, or any combination thereof.

[0035] The judgment data 13 is data including a judgment result when an expert judges the processing content to be performed on the earthquake motion time history data 11. The judgment data 13 may be editable via the input unit 43.

[0036] Specifically, the determination data 13 indicates at least one of a determination result when determining whether or not to perform a data exclusion process for excluding the seismic motion time history data 11 from the seismic motion data group when forming a seismic motion data group from a plurality of seismic motion time history data 11, and a determination result when determining whether or not to perform a noise removal process for removing a predetermined noise from the seismic motion time history data 11. In the example of FIG. 7, the above determination result is classified as either "necessary" or "unnecessary." Note that the determination result of the data removal process may be further classified according to the reason for removing the seismic motion time history data 11 from the seismic motion data group, for example, a low S / N ratio, a missing head portion, a missing tail portion, the presence of another earthquake or another event, and the like.

[0037] Furthermore, the determination data 13 further includes a determination result when at least one of a time condition for removing time history noise from the earthquake motion time history data 11 and a frequency condition for removing frequency noise from the earthquake motion time history data 11 is determined as a noise removal process. In the example of Fig. 7, the above-mentioned time condition includes at least one of a beginning time boundary that determines the time that is the boundary of the beginning part and an end time boundary that determines the time that is the boundary of the end part. Also, the above-mentioned frequency condition includes at least one of a low frequency boundary that determines the frequency that is the boundary of the low side and a high frequency boundary that determines the frequency that is the boundary of the high side.

[0038] (Regarding the acquisition process by the acquisition unit 411 and the learning data 14) 6, the acquisition unit 411 acquires a plurality of pieces of learning data 14, each of which is composed of case data 12 and judgment data 13, from the database 10. The learning data 14 is data used as teacher data (training data), verification data, and test data in supervised learning. A collection of the learning data 14, each of which is composed of a plurality of pieces of learning data 14, is called a learning dataset.

[0039] (Regarding the generation process by the generation unit 412 and the seismic motion data processing model 15) As shown in FIG. 6 , the generation unit 412 learns the correlation between the case data 12 and the judgment data 13 by machine learning based on the multiple learning data 14 acquired by the acquisition unit 411, thereby generating a seismic motion data processing model 15 as a learned model and storing it in the memory unit 40.

[0040] The earthquake motion data processing model 15 may be composed of a plurality of trained models. For example, the generating unit 412 may train a first trained model to learn a correlation between the case data 12 and a judgment result of a data elimination process in the judgment data 13, and may train a second trained model to learn a correlation between the case data 12 and a judgment result of a noise elimination process in the judgment data 13. The earthquake motion data processing model 15 may be generated for each utilization purpose of the earthquake motion time history data 11. In this case, for example, a database 10 may be prepared for each utilization purpose.

[0041] As described above, according to the earthquake motion data processing model generation device 4 and earthquake motion data processing model generation method of this embodiment, the knowledge gained by experts when determining the processing content to be performed on the earthquake motion time history data 11 can be incorporated into the earthquake motion data processing model 15 by machine learning, so that it is possible to generate a earthquake motion data processing model 15 that can determine appropriate processing content for the earthquake motion time history data 11 on behalf of experts.

[0042] (Configuration of the seismic motion data processing device 5 and the processes performed by each part) The earthquake motion data processing device 5 determines the processing content to be performed on the earthquake motion time history data 11a to be processed based on the earthquake motion data processing model 15 generated by the earthquake motion data processing model generation device 4, and processes the earthquake motion time history data 11a to be processed according to the processing content, thereby outputting the processed earthquake motion time history data 11b to an output medium (e.g., a storage medium, a display medium, a paper medium, etc.). The processing content (data exclusion processing and noise removal processing) for the earthquake motion time history data 11a to be processed is determined by the earthquake motion data processing device 5 using the earthquake motion data processing model 15 that has been trained by machine learning with the knowledge of experts, without being determined by an expert.

[0043] The seismic motion data processing device 5, like the seismic motion data processing model generation device 4, is composed of a general-purpose or dedicated computer, and as shown in FIG. 2, is equipped with a storage unit 50 composed of a HDD, memory, etc., a control unit 51 composed of a processor such as a CPU, GPU, etc., a communication unit 52 which is a communication interface with the network 6, an input unit 53 composed of a keyboard, mouse, etc., and a display unit 54 composed of a display, touch panel, etc.

[0044] The memory unit 50 stores a seismic motion data processing model 15 generated as a trained model by the seismic motion data processing model generation device 4, and a seismic motion data processing program 500 that controls the operation of the seismic motion data processing device 5 to realize a seismic motion data processing method.

[0045] The control unit 51 executes the seismic motion data processing program 500 to function as a reception unit 510, a determination unit 511, and a data processing unit 512. The functions of each unit will be described in detail later.

[0046] FIG. 8 is a functional explanatory diagram showing an example of a seismic motion data processing device 5 and a seismic motion data processing method according to an embodiment of the present invention.

[0047] (Regarding the reception process by the reception unit 510) The receiving unit 510 receives the earthquake motion time history data 11a to be processed. The earthquake motion time history data 11a to be processed is provided, for example, from the observation data providing device 2, and the processing content (data removal processing or noise removal processing) does not need to be determined by an expert. The receiving unit 510 may receive one piece of earthquake motion time history data 11a as the earthquake motion time history data 11a to be processed, or may receive a plurality of pieces of earthquake motion time history data 11a collectively. The receiving unit 510 may also receive the earthquake motion time history data 11a to be processed from a device other than the observation data providing device 2 (for example, the expert terminal device 3).

[0048] (Regarding the judgment process by judgment unit 511) The judgment unit 511 inputs case data 12 based on the earthquake motion time history data 11a to be processed that has been received by the receiving unit 510 into the earthquake motion data processing model 15, and judges the processing content to be performed on the earthquake motion time history data 11a to be processed, based on judgment data 13 output from the earthquake motion data processing model 15.

[0049] At that time, the determination unit 511 may convert the earthquake motion time history data 11a to be processed into the format of the case data 12 that can be input to the earthquake motion data processing model 15, and input the converted data to the earthquake motion data processing model 15. When the reception unit 510 receives a plurality of earthquake motion time history data 11 to be processed, the determination unit 511 may input each of the plurality of earthquake motion time history data 11 to the earthquake motion data processing model 15, and execute processing for each of the plurality of earthquake motion time history data 11 based on the determination data 13 output from the earthquake motion data processing model 15.

[0050] In addition, when the earthquake motion data processing model 15 is composed of a plurality of trained models, the case data 12 may be input to each of the trained models. For example, the determination unit 511 may perform the data removal process on the earthquake motion time history data 11a to be processed by inputting the case data 12 to a first trained model that has been trained to learn the correlation between the case data 12 and the determination result of the data removal process. In addition, the determination unit 511 may perform the noise removal process on the earthquake motion time history data 11a to be processed by inputting the case data 12 to a second trained model that has been trained to learn the correlation between the case data 12 and the determination result of the noise removal process.

[0051] (Regarding the output processing step by the data processing unit 512) The data processing unit 512 processes the earthquake motion time history data 11a to be processed according to the processing content determined by the determination unit 511, and outputs the processed earthquake motion time history data 11b to an output medium (e.g., a storage medium, a display medium, a paper medium, etc.).

[0052] For example, when the processing content determined by the determination unit 511 is to perform data removal processing, the data processing unit 512 removes the target earthquake motion time history data 11a from the earthquake motion data group. When the processing content determined by the determination unit 511 is to perform noise removal processing, the data processing unit 512 removes noise from the target earthquake motion time history data 11a. When a time condition is defined in the processing content, the data processing unit 512 removes time history noise based on the time condition, and when a frequency condition is defined in the processing content, the data processing unit 512 removes frequency noise based on the frequency condition.

[0053] When the output medium is a storage medium such as the storage unit 50, the data processing unit 512 stores and outputs the processed earthquake motion time history data 11b to the storage unit 50. When the output medium is a display medium such as the display unit 54, the data processing unit 512 generates data (output data) to be displayed on the display medium, and displays and outputs the data to the display unit 54. When the output medium is a paper medium, the data processing unit 512 generates data (output data) to be printed on the paper medium, and prints and outputs the data on the paper medium. The data processing unit 512 may provide the processed earthquake motion time history data 11b to the outside via the communication unit 52. Instead of or in addition to outputting the processed earthquake motion time history data 11b, the data processing unit 512 may output the determination data 13 for the earthquake motion time history data 11a to be processed.

[0054] As described above, according to the earthquake motion data processing device 5 and earthquake motion data processing method of this embodiment, the earthquake motion data processing model 15 that has learned the knowledge of experts is used to determine the processing content required for the earthquake motion time history data 11a to be processed, so that appropriate processing content can be performed on the earthquake motion time history data 11a to be processed instead of the experts.

[0055] (Other embodiments) Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and can be modified as appropriate without departing from the technical concept of the present invention.

[0056] In the above embodiment, the earthquake motion data processing model generating device 4 and the earthquake motion data processing device 5 are described as being configured as separate devices, but they may be configured as one device. Also, the expert terminal device 3 may be configured as at least one of the earthquake motion data processing model generating device 4 and the earthquake motion data processing device 5.

[0057] In the above embodiment, the earthquake motion data processing model generating program 400 and the earthquake motion data processing program 500 have been described as being stored in the storage units 40 and 50, respectively, but they may be provided by being recorded in a computer-readable recording medium such as a CD-ROM, DVD, or USB memory as a file in an installable or executable format. Also, the earthquake motion data processing model generating program 400 and the earthquake motion data processing program 500 may be provided by being stored on a computer connected to a network such as the Internet and downloaded via the network. [Explanation of symbols]

[0058] 1...Seismic motion time history data processing system, 2...Observation data providing device, 3...Expert terminal device, 4...seismic motion data processing model generating device, 5...seismic motion data processing device, 6...network, 10...Database, 11, 11a, 11b...Seismic motion time history data, 12...Case study data, 13...Judgment data, 14...Learning data, 15...Seismic motion data processing model, 40: memory unit, 41: control unit, 42: communication unit, 43: input unit, 44: display unit, 50: memory unit, 51: control unit, 52: communication unit, 53: input unit, 54: display unit, 400: Seismic motion data processing model generation program, 410...DB management section, 411...acquisition section, 412...generation section, 500…Seismic motion data processing program, 510: Reception unit, 511: Determination unit, 512: Data processing unit

Claims

1. A method for generating a seismic motion data processing model by machine learning using a computer to process seismic motion time history data, comprising: an acquisition step of acquiring a plurality of pieces of learning data composed of the case data and the judgment data from a database that stores case data based on the seismic motion time history data and judgment data including a judgment result when an expert judges a processing content to be performed on the seismic motion time history data in association with each other; and a generation step of generating the seismic motion data processing model as a trained model of the machine learning by learning a correlation between the case data and the judgment data by the machine learning based on the plurality of learning data acquired in the acquisition step, The determination data is A determination result when determining whether or not to perform a noise removal process for removing a predetermined noise from the seismic motion time history data is included. A method for generating earthquake motion data processing models.

2. The determination data is The noise removal process further includes at least one of a time condition when removing time history noise from the seismic motion time history data and a frequency condition when removing frequency noise from the seismic motion time history data. The earthquake motion data processing model generating method according to claim 1 .

3. The determination data is a determination result when determining whether or not to perform a data exclusion process for excluding the seismic motion time history data from the seismic motion data group when constructing the seismic motion data group from a plurality of the seismic motion time history data; The earthquake motion data processing model generating method according to claim 1 or 2.

4. A seismic motion data processing method for processing the seismic motion time history data based on the seismic motion data processing model generated by the seismic motion data processing model generation method according to any one of claims 1 to 3, comprising the steps of: A receiving step of receiving the seismic motion time history data to be processed; a determination step of determining the processing content to be performed on the earthquake motion time history data of the processing target based on the determination data output from the earthquake motion data processing model by inputting the case data based on the earthquake motion time history data of the processing target received in the receiving step into the earthquake motion data processing model, Seismic motion data processing methods.

5. A computer comprising: A control unit that executes each step included in the earthquake motion data processing model generation method according to any one of claims 1 to 3. Earthquake motion data processing model generation device.

6. A computer comprising: A control unit that executes each step included in the seismic motion data processing method according to claim 4. Seismic motion data processing device.

Images