Earthquake observation equipment, earthquake observation method, and recording medium

Abstract

An earthquake observation device includes: a waveform acquisition unit that acquires waveform data of a predetermined time period including the observation start time of a P wave; a delay time specification unit that inputs the waveform data into a training model and acquires the delay time from the observation start time of the P wave to the observation start time of an S wave from the training model; and an observation time estimation unit that estimates the observation start time of the S wave based on the observation start time of the P wave and the delay time.

Description

Technical Field

[0001] The present invention relates to earthquake observation equipment, an earthquake observation method and a recording medium. Background Art

[0002] When a large-scale earthquake occurs, it is expected that the occurrence of the earthquake will be reported as soon as possible. In response to such a demand, for example, in Japan, a system has been built that reports an earthquake warning through television broadcasting, mobile communications, etc. when a large-scale earthquake is determined to have occurred based on seismic waves.

[0003] As a related art, Patent Document 1 discloses a technique for realizing a B-Δ method for analyzing seismic waves through machine learning.

[0004] [Prior art literature]

[0005] [Patent Document]

[0006] [Patent Document 1] PCT International Patent Publication No. WO2018 / 008708 Summary of the Invention

[0007] Technical Problems to be Solved by the Invention

[0008] In order to report the occurrence of an earthquake as soon as possible, it is necessary to quickly specify the observation start time of the P wave (primary wave, pressure wave) and the observation start time of the S wave (secondary wave, shear wave) at the observation point of the seismic wave, and specify the source of the earthquake.

[0009] Therefore, there is a need for a technology that can quickly specify the observation start time of the S wave, which is technically difficult to specify.

[0010] An exemplary object of each exemplary aspect of the present invention is to provide an earthquake observation apparatus, an earthquake observation method, and a recording medium capable of solving the above-mentioned problems.

[0011] Means used to solve problems

[0012] According to an exemplary aspect of the present invention, an earthquake observation device includes: a waveform acquisition unit for acquiring waveform data of a predetermined time period including an observation start time of a P wave; a delay time specification unit for inputting the waveform data into a training model and acquiring a delay time from the observation start time of the P wave to the observation start time of the S wave from the training model; and an observation time estimation unit for estimating the observation start time of the S wave based on the observation start time of the P wave and the delay time.

[0013] According to an example aspect of the present invention, an earthquake observation method includes: acquiring waveform data of a predetermined time period including an observation start time of a P wave; inputting the waveform data into a training model, and acquiring a delay time from the observation start time of the P wave to the observation start time of the S wave from the training model; and estimating the observation start time of the S wave based on the observation start time of the P wave and the delay time.

[0014] According to an exemplary aspect of the present invention, a recording medium stores a program for causing a computer to execute: obtaining waveform data of a predetermined time period including an observation start time of a P wave; inputting the waveform data into a training model, and obtaining a delay time from the observation start time of the P wave to the observation start time of the S wave from the training model; and estimating the observation start time of the S wave based on the observation start time of the P wave and the delay time.

[0015] According to an exemplary aspect of the present invention, a recording medium stores a configuration program for configuring the following items as hardware: a waveform acquisition unit for acquiring waveform data of a predetermined time period including an observation start time of a P wave; a delay time specification unit for inputting the waveform data into a training model and acquiring a delay time from the observation start time of the P wave to the observation start time of the S wave from the training model; and an observation time estimation unit for estimating the observation start time of the S wave based on the observation start time of the P wave and the delay time.

[0016] Effects of the present invention

[0017] According to each exemplary aspect of the present invention, the observation start time of the S wave can be quickly specified. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a diagram showing an example of a configuration of an earthquake observation system according to an exemplary embodiment of the present invention.

[0019] Figure 2 is a diagram showing an example of a configuration of an earthquake detection apparatus according to an exemplary embodiment of the present invention.

[0020] Figure 3 is a diagram showing an example of a configuration of an earthquake observation apparatus according to an exemplary embodiment of the present invention.

[0021] Figure 4 is a diagram illustrating an example of seismic waves detected by a seismograph according to an exemplary embodiment of the present invention.

[0022] Figure 5 is a diagram illustrating an example of error distribution generated by an error distribution generating unit according to an exemplary embodiment of the present invention.

[0023] Figure 6 is a diagram illustrating an example of a processing flow of an earthquake observation apparatus according to an exemplary embodiment of the present invention.

[0024] Figure 7 is a diagram showing a configuration of an earthquake observation apparatus according to another exemplary embodiment of the present invention.

[0025] Figure 8 is a schematic block diagram illustrating a configuration of a computer according to at least one example embodiment. DETAILED DESCRIPTION

[0026] Hereinafter, example embodiments will be described in detail with reference to the accompanying drawings.

[0027] <Example Embodiment>

[0028] The earthquake observation system 1 according to an exemplary embodiment of the present invention is a system for estimating the observation start time of the S wave (the time when the observation of the S wave starts) by using a trained model. The trained model uses waveform data of a predetermined time period including the observation start time of the P wave contained in the earthquake wave as input, and learns the delay time from the observation start time of the P wave (the time when the observation of the P wave starts) to the observation start time of the S wave contained in the earthquake wave (the time when the observation of the S wave starts). Seismic waves are generated by earthquakes and represent the tremor at the observation point (for example, the installation location of the earthquake detection device 10 described below) in a time series when the tremor at the source of the earthquake propagates to the observation point. The P wave refers to the seismic wave that first arrives at the observation point when an earthquake occurs, and refers to a wave that represents the initial tremor. The S wave refers to the seismic wave that arrives at the observation point following the P wave when an earthquake occurs, and refers to a wave that exhibits a larger tremor called a large tremor.

[0029] like Figure 1 As shown, the earthquake observation system 1 includes an earthquake detection device 10 , an earthquake observation device 20 , and a earthquake analysis device 30 .

[0030] The earthquake detection device 10 detects earthquake waves. The earthquake observation device 20 measures P and S waves. The earthquake observation device 20 estimates the start time of S-wave observation using a training model that uses the earthquake waves detected by the earthquake detection device 10 as input and trains the delay time (time difference) from the start time of P-wave observation to the start time of S-wave observation. The earthquake analysis device 30 uses the measurement data of P and S waves measured by the earthquake observation device 20 to analyze earthquakes.

[0031] like Figure 2 As shown, the earthquake detection apparatus 10 includes a seismograph 101 and a communication unit 102 .

[0032] The seismograph 101 detects seismic waves.

[0033] The communication unit 102 communicates with the earthquake observation apparatus 20. For example, the communication unit 102 transmits the earthquake wave detected by the seismograph 101 to the earthquake observation apparatus 20.

[0034] like Figure 3 As shown, the earthquake observation device 20 includes: a P-wave observation time designation unit 201, a waveform acquisition unit 202, a model generation unit 203, an error distribution generation unit 204, a delay time designation unit 205, an S-wave observation time estimation unit (an example of an observation time estimation unit) 206, a search range determination unit 207, a earthquake analysis unit (an example of an observation time designation unit) 208 and a storage unit 209.

[0035] The P-wave observation time specification unit 201 specifies the observation start time of the P-wave by using the waveform data of the seismic wave acquired from the waveform acquisition unit 202 .

[0036] For example, the P-wave observation time specifying unit 201 determines whether the amplitude of the seismic wave indicated by the waveform data exceeds an amplitude of 10 times the noise level set in advance. The amplitude of 10 times the noise level described here is an example of a set value (threshold value). The set value can be set to any value.

[0037] When the P-wave observation time specifying unit 201 determines that the amplitude of the seismic wave indicated by the waveform data does not exceed an amplitude 10 times the preset noise level, the P-wave observation time specifying unit 201 performs the above-described determination operation again.

[0038] When the P-wave observation time designation unit 201 determines that the amplitude of the seismic wave indicated by the waveform data exceeds an amplitude 10 times the preset noise level, the P-wave observation time designation unit 201 specifies the time when the amplitude of the seismic wave indicated by the waveform data exceeded the preset noise level by backtracking from the time when this determination was performed. The time when the amplitude of the seismic wave indicated by the waveform data exceeded the preset noise level is the observation start time of the P wave.

[0039] The P-wave observation time designation unit 201 outputs the designated observation start time of the P-wave to the waveform acquisition unit 202 and the S-wave observation time estimation unit 206 .

[0040] As a method for the P-wave observation start time to be specified by the P-wave observation time specifying unit 201, any method may be used as long as the P-wave observation start time can be correctly specified.

[0041] The waveform acquisition unit 202 acquires waveform data of the seismic wave detected by the seismograph 101 from the earthquake detection apparatus 10. The waveform acquisition unit 202 outputs the acquired waveform data to the P-wave observation time specification unit 201.

[0042] The waveform acquisition unit 202 acquires waveform data of a predetermined period including the observation start time of the P wave detected by the seismograph 101 .

[0043] For example, the predetermined time period is one second before and one second after the observation start time of the P wave specified by the P wave observation time specifying unit 201 (ie, a total of 2 seconds). Figure 4 As shown, the waveform acquisition unit 202 specifies waveform data including waveform data of seismic waves one second before and one second after the observation start time of the P wave among the plurality of pieces of waveform data acquired from the earthquake detection apparatus 10 .

[0044] The waveform acquisition unit 202 outputs waveform data including waveform data of the seismic wave at the observation start time of the P wave to the delay time specification unit 205 .

[0045] The model generation unit 203 generates a training model by performing machine learning on a modeled neural network using a plurality of learning data. The neural network is, for example, a convolutional neural network including an input layer, an intermediate layer, and an output layer. Here, the learning data is data that associates waveform data for a predetermined period of time, including the start time of observation of the P wave, with the delay time (an example of the actual delay time) from the start time of observation of the P wave to the start time of observation of the S wave, obtained in advance through analysis, etc., in a one-to-one correspondence.

[0046] For example, the model generation unit 203 classifies multiple learning data into training data, evaluation data, and test data. The model generation unit 203 inputs the waveform data of the training data into the neural network. The neural network outputs the delay time from the observation start time of the P wave to the observation start time of the S wave. Each time the waveform data of the training data is input into the neural network and the delay time from the observation start time of the P wave to the observation start time of the S wave is output from the neural network, the model generation unit 203 changes the weights of the data coupling between nodes (i.e., changes the model of the neural network) in response to these outputs by performing backpropagation. The model generation unit 203 then inputs the waveform data of the evaluation data into the neural network of the model modified by the waveform data of the training data. The neural network outputs the delay time from the observation start time of the P wave to the observation start time of the S wave based on the input waveform data of the evaluation data. If necessary, the model generation unit 203 changes the weights of the data coupling between nodes based on the output of the neural network. The neural network generated by the model generation unit 203 in this manner is the training model. Then, as a final confirmation, the model generation unit 203 inputs the waveform data of the test data into the neural network of the trained model. The neural network of the training model outputs the delay time from the observation start time of the P wave to the observation start time of the S wave based on the waveform data of the input test data. When the delay time output by the neural network of the training model is within a predetermined error range relative to the delay time from the observation start time of the P wave to the observation start time of the S wave associated with the waveform data of the input test data for all test data, the model generation unit 203 determines that the neural network of the training model is the desired model. In addition, when the delay time output by the neural network of the training model is not within the predetermined error range relative to the delay time from the observation start time of the P wave to the observation start time of the S wave associated with the waveform data of the input test data for even one test data among the multiple test data, the model generation unit 203 generates a training model using new learning data.

[0047] The generation of the training model by the model generation unit 203 is repeatedly performed until the desired training model is obtained.

[0048] The model generation unit 203 records the generated training model in the storage unit 209 .

[0049] The error distribution generation unit 204 generates a frequency distribution of errors based on the neural network of the final training model and multiple learning data. The error is the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay from the observation start time of the P wave to the observation start time of the S wave.

[0050] For example, the error distribution generation unit 204 inputs the waveform data of the learning data into the neural network. The error distribution generation unit 204 obtains the output (delay time) of the neural network in response to the input waveform data. The error distribution generation unit 204 specifies the actual delay time from the observation start time of the P wave to the observation start time of the S wave associated with the waveform data in the learning data. The error distribution generation unit 204 subtracts the obtained output of the neural network from the specified actual delay time, thereby calculating the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave. The error distribution generation unit 204 calculates errors for multiple pieces of learning data. The error distribution generation unit 204 calculates the distribution of the frequency of occurrence of the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave, where the error is calculated for multiple pieces of learning data. The distribution of the occurrence frequency of the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave generated by the error distribution generation unit 204 is, for example, Figure 5 The distribution shown. Figure 5 The distribution shown is obtained by inputting waveform data of 25,000 learning data into the training model, and is the distribution of the frequency of occurrence of errors between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave. Figure 5 In FIG, the horizontal axis indicates time expressed with 1 scale being 1 / 100 seconds, and indicates the time of the error centered at error 0. In addition, in Figure 5 In , the vertical axis indicates the frequency (number of times) at which each error occurs when the waveform data of 25,000 pieces of learning data are input to the training model. Figure 5 In the example of the error distribution shown, the standard deviation σ is 164.18 (1.6418 seconds).

[0051] The error distribution generation unit 204 records the distribution of the occurrence frequency of the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave into the storage unit 209.

[0052] The delay time specification unit 205 acquires waveform data including waveform data of seismic waves at the observation start time of the P wave from the waveform acquisition unit 202 .

[0053] The delay time specifying unit 205 inputs the waveform data of the seismic wave at the observation start time of the P wave, which is included in the waveform data of the seismic wave acquired from the waveform acquiring unit 202, into the training model generated by the model generating unit 203. Here, the waveform data of the seismic wave input to the training model by the delay time specifying unit 205 is, for example, waveform data of one second before and one second after the observation start time of the P wave.

[0054] Furthermore, the delay time specifying unit 205 obtains the delay time from the P-wave observation start time to the S-wave observation start time output by the training model and outputs the obtained delay time from the P-wave observation start time to the S-wave observation time estimating unit 206.

[0055] For example, as disclosed in “Technical Reference Materials on Overview and Processing Methods of Earthquake Early Warning” [Online], December 13, 2016, Earthquake and Volcano Department, Japan Meteorological Agency, [searched on July 23, 2019], Internet (URL: https: / / www.data.jma.go.jp / svd / eew / data / nc / katsuyou / reference.pdf), the position of the earthquake source can be specified from the earthquake waves observed at one observation point by using the principal component analysis method and the B-Δ method,

[0056] PCT International Publication No. WO2018 / 008708, which is a patent document, discloses a configuration for implementing the B-Δ method through machine learning.

[0057] The above-mentioned processing performed by the delay time designation unit 205 is performed based on the B-Δ method, and is performed based on the correlation between the distance from the observation point to the earthquake source and the delay time from the observation start time of the P wave to the observation start time of the S wave (specifically, the shorter the distance from the observation point to the earthquake source, the shorter the delay time; and the longer the distance from the observation point to the earthquake source, the longer the delay time).

[0058] The S-wave observation time estimation unit 206 estimates the observation start time of the S-wave based on the observation start time of the P-wave and the delay time from the observation start time of the P-wave to the observation start time of the S-wave.

[0059] For example, the S-wave observation time estimation unit 206 acquires the observation start time of the P-wave from the P-wave observation time specification unit 201. Furthermore, the S-wave observation time estimation unit 206 acquires the delay time from the observation start time of the P-wave to the observation start time of the S-wave from the delay time specification unit 205. The S-wave observation time estimation unit 206 estimates the following time as the observation start time of the S-wave, that is, the time obtained by adding the delay time from the observation start time of the P-wave to the observation start time of the S-wave to the observation start time of the P-wave.

[0060] The S-wave observation time estimation unit 206 outputs the estimated observation start time of the S-wave to the search range determination unit 207. Furthermore, the S-wave observation time estimation unit 206 outputs the estimated observation start time of the S-wave to the earthquake analysis unit 208.

[0061] The search range determination unit 207 determines the search range of the S wave in the waveform data based on the following items: the distribution of the occurrence frequency of the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave; and the observation start time of the S wave estimated by the S wave observation time estimation unit 206.

[0062] For example, search range determination unit 207 obtains the S-wave observation start time estimated by S-wave observation time estimation unit 206 from S-wave observation time estimation unit 206. Search range determination unit 207 reads from storage unit 209 the distribution of the frequency of occurrence of errors between the delay time from the P-wave observation start time to the S-wave observation start time and the actual delay time from the P-wave observation start time to the S-wave observation start time, output by the training model. Search range determination unit 207 determines an error range (e.g., a time range of ±2σ) based on the distribution of the frequency of error occurrence. Here, σ represents the standard deviation. Search range determination unit 207 determines the error range based on the S-wave observation start time estimated by S-wave observation time estimation unit 206 (e.g., a time range of 2σ before and 2σ after the S-wave observation start time estimated by S-wave observation time estimation unit 206) as the search range for the S-wave in the waveform data.

[0063] The search range determination unit 207 records the determined search range of the S wave in the storage unit 209 .

[0064] The earthquake analysis unit 208 specifies the observation start time of the S wave in the waveform data based on the observation start time of the S wave estimated by the S wave observation time estimation unit 206 and the search range of the S wave determined by the search range determination unit 207 .

[0065] For example, the seismic analysis unit 208 acquires the observation start time of the S wave estimated by the S-wave observation time estimation unit 206 from the S-wave observation time estimation unit 206. Furthermore, the seismic analysis unit 208 acquires the search range of the S wave determined by the search range determination unit 207 from the storage unit 209. The seismic analysis unit 208 determines the observation start time of the S wave in the waveform data by measuring the S wave in the search range of the S wave determined by the search range determination unit 207 based on the observation start time of the S wave estimated by the S-wave observation time estimation unit 206 (an example of searching for the S wave).

[0066] Additionally, the seismic analysis unit 208 measures P waves.

[0067] For example, the measurement of the P wave and the S wave performed by the seismic analysis unit 208 can be performed by the measurement technology using AI disclosed in “[Automatic picking of P and S phases using a neural tree] Journal of Seismology (2006) 10: 39-63”.

[0068] The storage unit 209 stores various types of information necessary for processing performed by the earthquake observation apparatus 20 .

[0069] For example, the storage unit 209 stores a distribution of the frequency of occurrence of errors between the delay time from the start of observation of the P wave to the start of observation of the S wave, output by the training model, and the actual delay time from the start of observation of the P wave to the start of observation of the S wave, which distribution is generated by the error distribution generation unit 204. Furthermore, for example, the storage unit 209 stores learning data. Furthermore, for example, the storage unit 209 stores a training model generated by the model generation unit 203. Furthermore, for example, the storage unit 209 stores the search range for the S wave determined by the search range determination unit 207.

[0070] Next, a process of estimating the observation start time of the S wave by the seismic observation system 1 will be described.

[0071] Here, we will describe Figure 6 The processing flow of the earthquake observation system 1 is shown.

[0072] The following case will be described as an example. Model generation unit 203 generates a training model using the learning data stored in storage unit 209 and records the generated learning model in storage unit 209. Furthermore, search range determination unit 207 determines the search range for the S wave in the waveform data based on the distribution of the frequency of occurrence of errors between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave, and the S wave observation start time estimated by S-wave observation time estimation unit 206. Furthermore, storage unit 209 stores the search range for the S wave determined by search range determination unit 207.

[0073] The waveform acquisition unit 202 acquires waveform data of the seismic wave detected by the seismograph 101 from the earthquake detection apparatus 10 (step S1 ). The waveform acquisition unit 202 outputs the acquired waveform data to the P-wave observation time specification unit 201 .

[0074] The P-wave observation time specifying unit 201 acquires the waveform data of the seismic wave from the waveform acquiring unit 202. The P-wave observation time specifying unit 201 specifies the observation start time of the P-wave by using the acquired waveform data of the seismic wave (step S2). The P-wave observation time specifying unit 201 outputs the specified P-wave observation start time to the waveform acquiring unit 202 and the S-wave observation time estimating unit 206.

[0075] The waveform acquisition unit 202 acquires the P-wave observation start time specified by the P-wave observation time specification unit 201 from the P-wave observation time specification unit 201. The waveform acquisition unit 202 specifies waveform data of a predetermined time period including the acquired P-wave observation start time in the waveform data of the seismic wave detected by the seismograph 101 (step S3). The waveform acquisition unit 202 outputs the waveform data including the waveform data of the seismic wave at the P-wave observation start time to the delay time specification unit 205.

[0076] The delay time specifying unit 205 obtains waveform data, including the waveform data of the seismic wave at the observation start time of the P wave, from the waveform obtaining unit 202. The delay time specifying unit 205 inputs the waveform data of the seismic wave at the observation start time of the P wave included in the obtained waveform data of the seismic wave into the training model generated by the model generating unit 203. The delay time specifying unit 205 obtains the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model (step S4). The delay time specifying unit 205 outputs the obtained delay time from the observation start time of the P wave to the observation start time of the S wave to the S wave observation time estimating unit 206.

[0077] The S-wave observation time estimation unit 206 obtains the P-wave observation start time specified by the P-wave observation time specification unit 201 from the P-wave observation time specification unit 201. Furthermore, the S-wave observation time estimation unit 206 obtains the delay time from the P-wave observation start time to the S-wave observation start time from the delay time specification unit 205. The S-wave observation time estimation unit 206 estimates the S-wave observation start time based on the P-wave observation start time and the delay time from the P-wave observation start time to the S-wave observation start time (step S5). The S-wave observation time estimation unit 206 outputs the estimated S-wave observation start time to the earthquake analysis unit 208.

[0078] The seismic analysis unit 208 acquires the S-wave observation start time estimated by the S-wave observation time estimation unit 206 from the S-wave observation time estimation unit 206. Furthermore, the seismic analysis unit 208 acquires the S-wave search range determined by the search range determination unit 207 from the storage unit 209. The seismic analysis unit 208 specifies the S-wave observation start time by measuring the S-wave within a time range based on the acquired S-wave observation start time and the acquired S-wave search range (step S6).

[0079] Furthermore, the earthquake analysis unit 208 measures the P wave. The earthquake analysis unit 208 outputs measurement data obtained by measuring the P wave and the S wave to the earthquake analysis device 30.

[0080] The earthquake analysis apparatus 30 acquires the measurement data measured by the earthquake analysis unit 208 from the earthquake observation apparatus 20. The earthquake analysis apparatus 30 analyzes an earthquake by using the acquired measurement data.

[0081] The earthquake observation system 1 according to the exemplary embodiment of the present invention has been described above.

[0082] In the earthquake observation device 20 of the earthquake observation system 1, the waveform acquisition unit 202 acquires waveform data for a predetermined period of time, including the observation start time of the P wave. The delay time specification unit 205 inputs the waveform data into a training model and acquires the delay time from the observation start time of the P wave to the observation start time of the S wave from the training model. The S wave observation time estimation unit (an example of an observation time estimation unit) 206 estimates the observation start time of the S wave based on the observation start time of the P wave and the delay time.

[0083] In this manner, the earthquake observation system 1 can estimate the observation start time of the S wave from the P wave at the time when the S wave is not generated. Therefore, the earthquake observation system 1 can quickly specify the observation start time of the S wave.

[0084] The configuration of an earthquake observation apparatus 20 according to another exemplary embodiment of the present invention will be described.

[0085] like Figure 7 As shown, the earthquake observation apparatus 20 according to the present exemplary embodiment includes a waveform acquisition unit 202 , a delay time specification unit 205 , and an S-wave observation time estimation unit (observation time estimation unit) 206 .

[0086] The waveform acquisition unit 202 acquires waveform data of a predetermined period including the observation start time of the P wave.

[0087] The delay time specification unit 205 inputs the waveform data to the training model, and acquires the delay time from the observation start time of the P wave to the observation start time of the S wave from the training model.

[0088] The S-wave observation time estimation unit (observation time estimation unit) 206 estimates the observation start time of the S-wave based on the observation start time and the delay time of the P-wave.

[0089] The storage units, other storage devices, etc. in the exemplary embodiments of the present invention may be located anywhere as long as they can properly transmit and receive information. Alternatively, multiple storage units, other storage devices, etc. may be provided within a range where they can properly transmit and receive information, and data may be stored in a dispersed manner in these storage units, storage devices, etc.

[0090] In the process in the exemplary embodiment of the present invention, the order of the steps may be changed as long as appropriate steps are performed.

[0091] In an example embodiment of the present invention, an example of an earthquake observation device 20 including a search range determination unit 207 has been described, and the search range determination unit 207 determines the search range of the S wave in the waveform data based on the following items: the distribution of the occurrence frequency of the error between the delay time from the observation start time of the P wave to the observation start time of the S wave output by the training model and the actual delay time from the observation start time of the P wave to the observation start time of the S wave; and the observation start time of the S wave estimated by the S wave observation time estimation unit 206.

[0092] However, in another example embodiment of the present invention, the learning model may include a model for determining the search range of S waves in waveform data, and the model generation unit 203 may generate a training model including a model for determining the search range of S waves in waveform data by using the learning data.

[0093] In one exemplary embodiment of the present invention, the case has been described where the model generation unit 203 generates a training model as software and stores the training model in the storage unit 209 .

[0094] However, in another example embodiment of the present invention, the training model may be implemented as hardware.

[0095] For example, the model generation unit 203 may write a configuration program for implementing the processing performed by the training model stored in the storage unit 209 into programmable hardware such as an FPGA (Field Programmable Gate Array).

[0096] While the exemplary embodiment of the present invention has been described above, the earthquake observation system 1, earthquake detection device 10, earthquake observation device 20, earthquake analysis device 30, and other control devices described above may include a computer device. The aforementioned processing procedures are stored in the form of a program in a computer-readable recording medium, and the aforementioned processing is performed by a computer reading and executing the program. A specific example of a computer device will be described below.

[0097] Figure 8 is a schematic block diagram illustrating a configuration of a computer according to at least one example embodiment.

[0098] like Figure 8 As shown, the computer 5 includes a CPU 6, a main memory 7, a storage device 8 and an interface 9.

[0099] For example, each of the above-mentioned earthquake observation system 1, earthquake detection device 10, earthquake observation device 20, earthquake analysis device 30, and other control devices is installed in a computer 5. The operation of each of the above-mentioned processing units is stored in the form of a program in a storage device 8. The CPU 6 reads the program from the storage device 8, loads the program on the main memory 7, and executes the above-mentioned processing according to the program. According to the program, the CPU 6 ensures a storage area corresponding to each of the above-mentioned storage units in the main memory 7.

[0100] An HDD (Hard Disk Drive), an SSD (Solid State Drive), a magnetic disk, an optical magnetic disk, a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), a semiconductor memory, and the like are illustrative examples of the storage device 8. The storage device 8 may be an internal medium directly connected to a bus of the computer 5, or may be an external medium connected to the computer 5 via an interface 9 or a communication line. When the program is transmitted to the computer 5 via the communication line, the computer 5 receiving the transmitted program can load the program into the main memory 7 and execute the above-described processing. In at least one exemplary embodiment, the storage device 8 is a non-transitory tangible recording medium.

[0101] Furthermore, the above-mentioned program can realize some of the above-mentioned functions. Furthermore, the program can be a file that can realize the above-mentioned functions in combination with a program already recorded on a computer device, ie, a so-called difference file (difference program).

[0102] Although some exemplary embodiments of the present invention have been described, these exemplary embodiments are examples and do not limit the scope of the present invention. Various additions, omissions, substitutions and changes can be made to the exemplary embodiments without departing from the spirit of the present invention.

[0103] This application is based upon and claims the benefit of priority from Japanese patent application No. 2019-150624, filed on August 20, 2019, the disclosure of which is incorporated herein in its entirety by reference.

[0104] Industrial Applicability

[0105] The present invention can be applied to earthquake observation equipment, earthquake observation methods, and recording media.

[0106] Description of Reference Numerals

[0107] 1: Earthquake Observation System

[0108] 5: Computer

[0109] 6: CPU

[0110] 7: Main memory

[0111] 8: Storage devices

[0112] 9: Interface

[0113] 10: Earthquake detection equipment

[0114] 20: Earthquake observation equipment

[0115] 30: Seismic analysis equipment

[0116] 101: Seismograph

[0117] 102: Communication unit (communication device)

[0118] 201: P-wave observation time designation unit (P-wave observation time designation device)

[0119] 202: Waveform acquisition unit (waveform acquisition device)

[0120] 203: Model generation unit (model generation device)

[0121] 204: Error distribution generating unit (error distribution generating means)

[0122] 205: Delay time specifying unit (delay time specifying device)

[0123] 206: S-wave observation time estimation unit (observation time estimation unit, observation time estimation device)

[0124] 207: Search range determination unit (search range determination means)

[0125] 208: Earthquake Analysis Unit (Earthquake Analysis Device)

[0126] 209: Storage unit (storage device).

Claims

1. An earthquake observation device comprising: a waveform acquisition unit, configured to acquire waveform data of a predetermined time period including a start time of observation of the P wave; a delay time specifying unit for inputting the waveform data into a training model and acquiring a delay time from the observation start time of the P wave to the observation start time of the S wave from the training model; an observation time estimating unit, configured to estimate the observation start time of the S wave based on the observation start time of the P wave and the delay time; a search range determination unit for determining a search range for the S wave in the waveform data based on: a distribution of frequencies of errors between the delay time output by the training model and an actual delay time; and an estimated observation start time of the S wave; and An observation time specifying unit is configured to specify an observation start time of the S wave by searching for the S wave within the search range.

2. A method for earthquake observation, comprising: Acquiring waveform data for a predetermined time period including a start time of observation of the P wave; inputting the waveform data into a training model, and acquiring a delay time from an observation start time of the P wave to an observation start time of the S wave from the training model; estimating the S-wave observation start time based on the P-wave observation start time and the delay time; determining a search range for the S wave in the waveform data based on: a distribution of frequencies of errors between the delay time output by the training model and an actual delay time; and an estimated observation start time of the S wave; and The observation start time of the S wave is specified by searching for the S wave within the search range.

3. A recording medium storing a program for causing a computer to perform the following operations: Acquiring waveform data for a predetermined time period including a start time of observation of the P wave; inputting the waveform data into a training model, and acquiring a delay time from an observation start time of the P wave to an observation start time of the S wave from the training model; estimating the S-wave observation start time based on the P-wave observation start time and the delay time; determining a search range for the S wave in the waveform data based on: a distribution of frequencies of errors between the delay time output by the training model and an actual delay time; and an estimated observation start time of the S wave; and The observation start time of the S wave is specified by searching for the S wave within the search range.

Images