[0001]The present invention, for estimating the epicenter distance when the earthquake occurred, relates epicentral distance estimation apparatus and epicentral distance estimation method further relates to computer-readable recording medium a program for realizing these.
BACKGROUND
[0002]In the event of an earthquake, in order to estimate the arrival time of seismic intensity and principal motion at around quickly it is necessary to identify the epicenter distance. Usually, the depth of the epicenter distance is specified based on the seismic intensity detected by the seismograph in multiple locations.
[0003]
　However, the epicenter of the position, or a submarine, when the installation density of the seismic intensity meter or a low area, become a situation in which the time to acquire the seismic intensity, which is measured at a plurality of seismic intensity meter too much, the epicentral distance specific There would be delayed. Therefore, in recent years, only using the seismic intensity measured by seismograph alone, techniques for identifying the epicenter distance has been developed.
[0004]
　As such a technology, "the rise of the strength of the seismic waveform data at the time of the earthquake of reach, becomes stronger as there is an earthquake close to the epicenter, epicenter becomes gentle as a distant earthquake," taking advantage of the fact, technique of estimating the epicenter distance is known (e.g., see Patent Document 1).
[0005]
　Specifically, in the technique disclosed in Patent Document 1, the absolute value of the time series data obtained from seismometers y (t), the time t, the time seismometer detects earthquake as t = 0, waveform of the acquired seismic initial part seismometer is fitting the function indicated by the number 1 below. In Equation 1 below, A is a parameter related to the maximum amplitude of the initial portion, B is a parameter related to time variation of the initial amplitude of the seismic wave. It should be noted that, in fact, in the fitting, relying on the human experience and intuition, is a complex individual point attribute, what if you want to influence the work to be reflected in the unknown parameters A and B is performed .
[0006]
[Number 1]

[0007]
　Then, in the technique disclosed in Patent Document 1, the least square method, it is required parameters A and B. Among them, there is a correlation between the parameters B and epicenter distance, this correlation has been found to be unaffected in magnitude. Therefore, in advance formulate a correlation between the parameter B and the epicenter distance, from a waveform shape of the seismic initial portion, by calculating a parameter B using equation 1, the epicenter distance is specified. According to the technique disclosed in Patent Document 1, the waveform of the seismic initial portion, can quickly identify epicenter distance.
CITATION
Patent Document
[0008]
Patent Document 1: JP 2002-277557 JP
Summary of the Invention
Problems that the Invention is to Solve
[0009]
　However, it was the technique disclosed in Patent Document 1, in some circumstances, may live a situation that can not be calculated coefficients A and B are generated, there is a problem that reliability is not sufficient. Further, the disclosed in Patent Document 1 technology, also a problem that it is difficult shorten the time required for the calculation of the epicenter distance.
[0010]
　An example of an object of the present invention is to solve the above problems, it can for calculating the epicenter distance stably, and give work to shorten the calculation time, epicentral distance estimation apparatus, epicentral distance estimation method, and the computer and to provide a readable recording medium.
Means for Solving the Problems
[0011]
　To achieve the above object, epicentral distance estimation apparatus according to an aspect of the present invention,
　obtains the waveform data of the generated seismic, and earthquake information acquisition unit,
　to learn the relationship between the waveform data and the epicentral distance seismic give the obtained learning model, by applying the obtained the waveform data, to estimate the epicenter distance, and estimation processing unit,
and a, characterized in that.
[0012]
　In order to achieve the above object, epicentral distance estimation method in one aspect of the present invention,
(a) acquiring the waveform data of the generated seismic, a step,
(b) the relationship between the waveform data and the epicentral distance Earthquake learning model obtained by learning, by applying acquired the waveform data, to estimate the epicenter distance, a step,
having, characterized in that.
[0013]
　Furthermore, in order to achieve the above object, a computer-readable recording medium according to an aspect of the present invention,
the computer
obtains the waveform data of the earthquake (a), a step,
and (b) seismic waveform data a learning model obtained by learning the relation between the epicenter distance, by applying acquired the waveform data, to estimate the epicenter distance, a step,
thereby executing including instructions, which records a program the features.
Effect of the invention
[0014]
　As described above, according to the present invention, it is possible to calculate the epicenter distance stably, and it is possible to shorten the calculation time.
BRIEF DESCRIPTION OF THE DRAWINGS
[0015]
[1] Figure 1 is a block diagram showing a schematic configuration of epicentral distance estimation apparatus according to the first embodiment of the present invention.
FIG. 2 is a block diagram concretely showing the configuration of the epicenter distance estimation apparatus according to the first embodiment of the present invention.
FIG. 3 is a diagram showing an example of input data and correct answer data used for learning in the first embodiment.
[4] FIG 4 is a flow diagram illustrating the operation of the learning process executed in the epicentral distance estimation apparatus according to the first embodiment of the present invention.
FIG. 5 is a flow diagram showing the operation at the time estimation process executed in the epicentral distance estimation apparatus according to the first embodiment of the present invention.
FIG. 6 is a block diagram concretely showing the configuration of the epicenter distance estimation apparatus according to the second embodiment of the present invention.
[7] FIG. 7 is a flow diagram illustrating the operation of the learning process executed in the epicentral distance estimation apparatus according to the second embodiment of the present invention.
[8] FIG. 8 is a flow diagram illustrating the estimation process runtime behavior of epicentral distance estimation apparatus according to the second embodiment of the present invention.
[9] FIG. 9 is a block diagram showing an example of a computer that realizes the epicenter distance estimation apparatus according to Embodiment 1 and 2 of the present invention.
DESCRIPTION OF THE INVENTION
[0016]
(Embodiment 1)
　Hereinafter, in the first embodiment of the present invention, epicentral distance estimation apparatus, epicentral distance estimation method, and the program will be described with reference to FIGS.
[0017]
[Device Configuration]
　First, with reference to FIG. 1, it will be described a schematic configuration of epicentral distance estimation apparatus according to the first embodiment. Figure 1 is a block diagram showing a schematic configuration of epicentral distance estimation apparatus according to the first embodiment of the present invention.
[0018]
　1, epicentral distance estimation apparatus 10 according to the first embodiment is an apparatus for estimating the epicenter distance from the waveform data measured during an earthquake. As shown in FIG. 1, epicenter distance estimation apparatus 10 includes a seismic data acquisition unit 11, the estimated processing unit 12.
[0019]
　Earthquake information acquiring unit 11 acquires the waveform data of the generated seismic. Estimation processing unit 12, a learning model, by applying the waveform data acquired by the seismic information acquisition unit 11, estimates the epicenter distance. Learning model is previously obtained by learning the relationship between waveform data and epicentral distance earthquake.
[0020]
　Thus, in the first embodiment, unlike the conventional, without fitting the waveform data to a function, since the epicenter distance is estimated, it is possible to calculate the epicenter distance stably. In the first embodiment, since there is no need calculation processing by the least square method, so that also reduced shortening of the calculation time.
[0021]
　Subsequently, with reference to FIG. 2, further specifically describes the structure of the epicentral distance estimation apparatus according to the first embodiment. Figure 2 is a block diagram concretely showing the configuration of the epicenter distance estimation apparatus according to the first embodiment of the present invention.
[0022]
　As shown in FIG. 2, in the first embodiment, epicenter distance estimation apparatus 10, via the network, and is connected to the seismic sensing device 20 and the seismic activities overall monitoring system 30. Of these, earthquake detection device 20 includes a seismometer, the seismic waves by seismometer is detected, it transmits the waveform data of the detected seismic epicenter distance estimation apparatus 10. In the first embodiment, earthquake detection device 20, the acquisition source of the waveform data of earthquake information acquisition unit 11.
[0023]
　In the example of FIG. 2, only a single seismic detector 20 is illustrated, the number of seismic sensing device 20 epicentral distance estimation apparatus 10 is connected is not particularly limited. However, seismic sensing device 20 as the acquisition source of the earthquake information acquisition unit 11 may be any one of these.
[0024]
　Seismic activity, such as a comprehensive monitoring system 30 is a system that held by the Japan Meteorological Agency in Japan, an earthquake occurs, to calculate the Japan Meteorological Agency magnitude, based on the calculated Meteorological Agency magnitude, to predict a tsunami height. In addition, seismic activity, such as a comprehensive monitoring system 30, and the tsunami height calculated Meteorological Agency magnitude and predicted, in various media, to deliver as earthquake early warning.
[0025]
　In the first embodiment, epicentral distance estimation apparatus 10, the epicenter distance estimated, input to seismic activities overall monitoring system 30. Thus, seismic activity, etc. General monitoring system 30 uses the epicenter distance estimated by epicentral distance estimation apparatus 10, executes the prediction of JMA magnitude calculation and tsunami height.
[0026]
　Further, as shown in FIG. 2, in the first embodiment, epicentral distance estimation apparatus 10 includes, in addition to earthquake information acquisition unit 11 and the estimation processor 12 described above, the learning information acquisition unit 13, a learning section 14 and a storage unit 15. Incidentally, FIG. 2 shows an example of epicentral distance estimation apparatus 10, the learning information acquisition unit 13, the learning unit 14 and memory unit 15 may be provided in a device other than epicentral distance estimation apparatus 10.
[0027]
　Learning information acquisition unit 13 acquires the waveform data to be input in the training of the learning section 14 to be described later, the epicenter distance similarly the correct data in the learning, and inputs them to the learning unit 14. The acquisition source of the input data and the correct answer data are not particularly limited.
[0028]
　Learning unit 14, the waveform data of earthquakes as input data, the epicenter distance seismic as the correct data, to learn the relationship between the waveform data and the epicentral distance, to create a learning model 16 showing the learning result. Also, the learning unit 14 stores the generated training model 16 in the storage unit 15.
[0029]
　Figure 3 is a diagram showing an example of input data and correct answer data used learned in the first embodiment. Figure 3 is epicentral distance are shown different waveform data. Each waveform data shown in FIG. 3 is a waveform data observed in the past earthquakes. Further, epicenter distance corresponding to the waveform data is correct data. Learning section 14 as input data of each waveform data shown in FIG. 3, the epicenter distance as the correct data and performs learning.
[0030]
　Further, in the first embodiment, the correct answer data can be used data JMA is published. The data JMA is published, epicenter distance of each observation point, includes a seismic source elements, they JMA was calculated by centralized systems gage value (http: //www.data.jma.go .jp / svd / eqev / data / bulletin / deck.html) has been determined from. Further, in the first embodiment, the input data and accuracy data used for learning, and even better in which seismic intensity is calculated from the seismic is the set value (e.g., seismic intensity 4) or more.
[0031]
　In the first embodiment, the learning unit 14, for example, by machine learning to build a neural network can be a neural network and the learning model 16. Specifically, the learning unit 14, an input layer, an intermediate layer, and the hierarchical neural network with the output layer, using the input data and the correct answer data, the connection weight between the nodes layers adjacent by adjusting the value, to generate a learning model.
[0032]
　Further, in the first embodiment, by the learning section 14 "learning" means so-called "machine learning". Further, by the learning section 14 "learning" is not limited to the deep learning using a neural network described above, learning using logistic regression, learning using service port vector machine learning using a decision tree, with heterogeneous learning etc. it may be.
[0033]
　Earthquake information acquisition unit 11 in the first embodiment, a single seismic sensing device 20, receives the waveform data of the generated seismic. Further, earthquake information acquiring unit 11 sends the received waveform data to the estimation processing unit 12.
[0034]
　Estimation processing unit 12, in the first embodiment, by accessing the storage unit 15, acquires the learning model 16, the learning model 16 acquired, applies the waveform data sent from the learning information acquisition unit 13 it is, to estimate the epicenter distance.
[0035]
　In the first embodiment, the learning section 14, as the correct data, further, using the focal depth of an earthquake, by learning waveform data, the epicenter distance and focal and depth, the relationships, learning model 16 can also be generated. In this case, the estimation processing unit 12, in addition to the epicenter distance, also estimated focal depth.
[0036]
　Furthermore, in the first embodiment, the learning section 14, in addition to the waveform data, even point data of the point where the waveform data obtained can be used as input data. In this case, the learning unit 14, the waveform data and location data, to learn the relationship between the epicenter distance (or epicenter distance and focal depth), to generate a learning model 16.
[0037]
　Here, the point at which the waveform data obtained, a point at which seismic waves was the source of the waveform data was observed. As the spot data, for example, surface soil amplification factor, data indicating the state of the plate, the data indicating the volcano are present near the point where the seismic wave is observed, crust thickness, lithosphere thickness, and the like. Thus, as input data, if generating a training model 16 with two waveform data and the location data, improve the accuracy of the estimation process is achieved.
[0038]
　If the point data is used as input data for learning, earthquake information acquiring unit 11, in addition to the waveform data of the generated seismic, location data of the point where the waveform data is obtained, i.e., earthquake detection device 20 but also to get the point data of the point in which it is installed.
[0039]
　Also, the point data in advance, for each seismic detector 20 may be stored in the storage unit 15, in this embodiment, earthquake information acquisition unit 11, each time for acquiring waveform data from the storage unit 15, to obtain the corresponding point data. Also, the point data from earthquake detection device 20 may be sent along with the waveform data, in this embodiment, earthquake information acquiring unit 11 acquires the location data together with the waveform data.
[0040]
　Further, if the location data is used as input data for the learning, the estimation processing unit 12, a learning model 16 generated by the learning unit 14, by applying the waveform data and the location data obtained, epicentral distance (or epicenter distance and focal depth) to estimate.
[0041]
　In the first embodiment, the input data and correct answer data are not intended to be limited to the above example. As input data, other than the waveform data and location data may be used. Also, as the correct data, other than epicentral distance and focal depth it may be used.
[0042]
[Device Operation]
　Next, the operation of the epicentral distance estimation apparatus 10 according to Embodiment 1 will be described with reference to FIGS. In the following description, it is referred to appropriately FIGS. 1 to 3. In the first embodiment, by operating the epicenter distance estimation apparatus 10, epicentral distance estimation method is implemented. Therefore, description of epicentral distance estimation method in the first embodiment, substitute for explaining the operation of the epicentral distance estimation apparatus 10 below.
[0043]
　In the first embodiment, epicentral distance estimation apparatus 10 mainly performs the learning process and the estimation process. First, a description will be given of the learning process. Figure 4 is a flow diagram illustrating a learning process runtime behavior of epicentral distance estimation apparatus according to the first embodiment of the present invention.
[0044]
　As shown in FIG. 4, first, the learning information acquisition unit 13 acquires the input data and correct answer data (step A1). Specifically, in step A1, the learning information acquisition unit 13, as input data, in addition to the waveform data, location data is also acquired, as the correct data, in addition to the epicenter distance, also acquires focal depth.
[0045]
　Then, the learning unit 14 already determines whether learning model 16 is present (step A2). Specifically, the learning unit 14 determines whether the learning model 16 is stored in the storage unit 15.
[0046]
　It is determined in step A2, if the learning model 16 does not exist yet, the learning section 14 learns the waveform data and location data, the relationship between the epicenter distance and focal depth, learning model showing the learning result 16 newly generates (step A3).
[0047]
　Specifically, in step A3, the learning unit 14, the learning to build a neural network and this learning model 16. Also, the learning unit 14 stores a learning model 16 created in the storage unit 15.
[0048]
　On the other hand, the result of the determination in step A2, if the learning model 16 already exists, the learning unit 14 uses the correct data and the input data obtained in step A1, and updates the existing training model 16 (step A4). Specifically, the learning unit 14 uses the the acquired input data and correct answer data in the step A1, updates the value of the connection weight between the nodes.
[0049]
　By executing the steps A1 ~ A4, learning model is created or updated is performed. Then, using a learning model created or updated, the estimation process is executed. Figure 5 is a flow diagram illustrating the estimation process runtime behavior of epicentral distance estimation apparatus according to the first embodiment of the present invention.
[0050]
　As shown in FIG. 5, first, earthquake information acquiring unit 11, from the earthquake detection device 20, the waveform data of the generated seismic is transmitted, it receives the waveform data transmitted (step B1).
[0051]
　Next, earthquake information acquiring unit 11 from the storage unit 15, acquires the location data of the point where the seismic sensing device 20 that has transmitted the waveform data is installed (step B2). Incidentally, the point data, if you are sent along with the waveform data, earthquake information acquiring unit 11 receives the location data transmitted.
[0052]
　Next, the estimation processing unit 12, a waveform data received at step B1, the spot data acquired in step B2, is applied to the learning model 16 created or updated by the learning process shown in FIG. 4 estimates the epicentral distance and focal depth (step B3).
[0053]
　By executing the steps B1 ~ B3, on the basis of the waveform data obtained from a single seismic detector 20, so that the epicentral distance and focal depth is estimated.
[0054]
Effects of First Embodiment
　According to the first embodiment as described above, without fitting the waveform data to a function, from a single waveform data, epicentral distance and focal depth is estimated. Further, estimation process, because performed by the learning model 16, epicentral distance and focal depth are stable, and will be calculated in a short time.
[0055]
　In other words, in the first embodiment, unlike the conventional method disclosed in Patent Document 1, by hand, unnecessary work of complex individual point attributes, what is how to influence is reflected in the unknown parameters become. According to the first embodiment, it is possible to obtain the available precision of the information only by objective waveform data and machine learning. Epicentral distance estimation apparatus 10 in the first embodiment, many points, it is possible to introduce into many regions.
[0056]
　As described above, in this embodiment, also be estimated focal depth by a single waveform data in the above was the technique disclosed in Patent Document 1, the estimation of the focal depth is not possible. When using the technique disclosed in Patent Document 1, for focal depth of the measurement, it is necessary to measure the results in a plurality of seismometers.
[0057]
[Modification 1]
　The following will describe a modification example of the first embodiment. First, in the first modification, the learning unit 14, for each quantity waveform that is set to generate a learning model 16. Specifically, the waveform amount is expressed by the elapsed time from the time of earthquake. Thus, the learning section 14, the waveform data acquired by the learning information acquisition unit 13, for each set elapsed time, cut by an amount of the waveform data of the elapsed time, as input data the waveform data cut out, by performing learning to generate a learning model 16. Thus, learning model 16 is generated for each waveform amount.
[0058]
　Further, in the first modification, the estimation processing unit 12 calculates the waveform of the waveform data of the generated seismic, based on the calculated waveform amount, from among a plurality of learning models 16 being generated, using learning to select a model. The estimation processing unit 12, a learning model 16 chosen by applying the waveform data of the generated seismic estimates the epicenter distance (or epicenter distance and focal depth).
[0059]
　In general, the estimation of epicentral distance and focal depth for the earthquake early warning, even if the waveform of the waveform data is smaller is determined. Therefore, the waveform data acquired by the seismic information acquisition unit 11 is not limited to be constant, the waveform of the waveform data used for the generation of learning models, and the waveform of the waveform data in the generated earthquake coincide not, there is a possibility that the estimation accuracy decreases. However, according to the modified example 1, depending on the waveform of the waveform data in the generated seismic, since learning model 16 is selected, so that the reduction of the above-described estimation accuracy is avoided.
[0060]
[Modification 2]
　In Modification 2, the learning unit 14, for each observation point of the waveform data to be input data, and generates the learning model 16. Specifically, the learning unit 14, for each seismic detector (per seismometers), where using only acquired waveform data to generate learning model 16.
[0061]
　Further, in the modified example 2, the estimation processing unit 12, the observation point of the waveform data of the generated seismic (i.e., earthquake detection device 20 that is the source of the waveform data) to identify, based on the specified observation point , from among a plurality of learning models 16 are generated, selects the learning model to be used. The estimation processing unit 12, a learning model 16 chosen by applying the waveform data of the generated seismic estimates the epicenter distance (or epicenter distance and focal depth).
[0062]
　According to the second modification, characteristics of each observation point, i.e., without performing learning using the location data, estimation processing the combined becomes possible to the characteristics of the observation point. In the observation point can not be sufficient input data, since the sufficient learning becomes difficult, formation of learning models for such observation point is difficult.
[0063]
[Modification 3]
　In Modification 3, the learning unit 14, for each soil characteristics at the observation point of the waveform data to be input data, and generates the learning model 16. Specifically, for example, depending on the soil characteristics such as site amplification factor (the value of the point data), the observation point (seismic sensing device 20) are grouped. In this case, the learning unit 14, for each group, using only the waveform data obtained by the group, it generates the learning model 16.
[0064]
　Further, in the modified example 3, the estimation processing unit 12 identifies the soil characteristics of the observation point of the waveform data of the generated seismic, based on the specified ground characteristics, from a plurality of learning models 16 being generated, to select a learning model to be used. The estimation processing unit 12, a learning model 16 chosen by applying the waveform data of the generated seismic estimates the epicenter distance (or epicenter distance and focal depth).
[0065]
　According to the third modification, even if the observation point input data is not sufficiently ensured was present, without performing learning using location data, estimation processing the combined becomes possible to the characteristics of the observation point .
[0066]
Programs
　program in the first embodiment, the computer, the step A1 ~ A4 shown in FIG. 4, may be a program for executing the steps B1 ~ B3 shown in FIG. To install the program on a computer, by executing, it is possible to realize the epicentral distance estimating device 10 in Embodiment 1 and epicentral distance estimation method. In this case, the computer of a CPU (Central Processing Unit), earthquake information acquiring unit 11, the estimation processing unit 12, the learning information acquisition unit 13, and functions as a learning unit 14 performs processing.
[0067]
　The program in the first embodiment may be executed by a computer system built by a plurality of computers. In this case, each computer, respectively, earthquake information acquiring unit 11, the estimation processing unit 12, the learning information acquisition unit 13, and may function as either a learning unit 14. The storage unit 15 may be built on a different computer from the computer that executes the program of the present embodiment.
[0068]
(Embodiment 2)
　Next, in the second embodiment of the present invention, epicentral distance estimation apparatus, epicentral distance estimation method, and the program will be described with reference to FIGS.
[0069]
[Device Configuration]
　First, with reference to FIG. 6, the configuration of the epicentral distance estimation apparatus according to the second embodiment. Figure 6 is a block diagram concretely showing the configuration of the epicenter distance estimation apparatus according to the second embodiment of the present invention.
[0070]
　As shown in FIG. 6, epicentral distance estimation apparatus 40 according to the second embodiment includes a waveform pre-processing section 41, in this respect, the epicenter distance estimation in the first embodiment shown in FIGS. 1 and 2 It is different from the device 10. The following description focuses on differences from the first embodiment.
[0071]
　Waveform preprocessing section 41 performs preprocessing on waveform data acquired in the waveform data and seismic information acquisition unit 11, used in the learning section 14 as input data. The pre-processing, image conversion processing, envelope conversion process, the bandpass conversion process, a differential conversion, and Fourier transform treatment.
[0072]
　Specifically, the image conversion processing, the waveform data is a process of converting the image data of an image to display it graphically. According to the image conversion processing, the learning unit 14, for performing learning based on the image data, is considered to learning process is facilitated.
[0073]
　Further, the envelope conversion process is a process for the gentle waveform of the waveform data. If you run enveloping, since it is easy to identify the rising characteristics of the seismic waves, learning model 16 rising characteristics of the seismic wave is reflected is generated.
[0074]
　Bandpass conversion process is a highlight processing the waveform of the specific period. According to the band pass converting process, since the characteristics of the seismic wave is emphasized, learning model 16, wherein the seismic wave is reflected is generated.
[0075]
　Further, the differential conversion process, by differentiating the waveform data, a process for converting the acceleration data. May be due to differential conversion, because it is easy to identify the rising characteristics of the seismic waves, learning model 16 rising characteristics of the seismic wave is reflected is generated.
[0076]
　Further, the Fourier transform processing is processing for obtaining the frequency distribution of the waveform data. According to Fourier transform, the difference in the period of each waveform data is emphasized, learning model 16 is the period of the seismic wave is reflected is generated.
[0077]
　Waveform pre-processing section 41, image conversion processing, envelope conversion process, the bandpass conversion processing may be performed differential conversion, and any one of the Fourier transform process, or two or more.
[0078]
[Device Operation]
　Next, the operation of the epicentral distance estimation apparatus 40 according to the second embodiment will be described with reference to FIGS. In the following description, it is referred to FIG. 1 as needed to 6. In the second embodiment, by operating the epicenter distance estimation apparatus 40, epicentral distance estimation method is implemented. Therefore, description of epicentral distance estimation method in the second embodiment, replace the description of the operation of the following epicentral distance estimation apparatus 40.
[0079]
　First, a description will be given of the learning process. Figure 7 is a flow diagram illustrating the operation of the learning process executed in the epicentral distance estimation apparatus according to the second embodiment of the present invention.
[0080]
　7, first, the learning information acquisition unit 13 acquires the input data and correct answer data (step A11). Further, the learning information acquisition unit 13 inputs the obtained data to the waveform pre-processing section 41.
[0081]
　Next, the waveform pre-processor 41 executes pre-processing for the waveform data included in the input data acquired in step A11 (step A12). Then, the waveform pre-processor 41 includes a pre-processed waveform data, and other input data (point data), the correct answer data are inputted to the learning section 14.
[0082]
　Then, the learning unit 14 already determines whether learning model 16 is present (step A13).
[0083]
　It is determined in step A13, if the learning model 16 does not exist yet, the learning section 14 learns the waveform data and location data, the relationship between the epicenter distance and focal depth, learning model showing the learning result 16 newly generates (step A14).
[0084]
　On the other hand, the result of the determination in step A13, if the learning model 16 already exists, the learning section 14 uses the input data and correct answer data, and updates the existing learning model 16 (step A15).
[0085]
　By executing the steps A11 ~ A15, creating or updating a training model 16 is performed. Then, using a learning model 16 created or updated, the estimation process is executed. Figure 8 is a flow diagram illustrating the estimated processing execution time of the operation of the epicentral distance estimation apparatus according to the second embodiment of the present invention.
[0086]
　As shown in FIG. 8, first, earthquake information acquiring unit 11, from the earthquake detection device 20, the waveform data of the generated seismic is transmitted, it receives the waveform data transmitted (step B11).
[0087]
　Next, earthquake information acquiring unit 11 from the storage unit 15, acquires the location data of the point where the seismic sensing device 20 that has transmitted the waveform data is installed (step B12). Incidentally, the point data, if you are sent along with the waveform data, earthquake information acquiring unit 11 receives the location data transmitted.
[0088]
　Next, the waveform pre-processor 41 performs pre-processing on the waveform data received at step B11 (step B13). Then, the waveform pre-processor 41, before the waveform data after processing, and a spot data, and inputs to the estimation processor 12.
[0089]
　Next, the estimation processing unit 12 applies the waveform data after pretreatment with step B13, the spot data acquired in step B12, the learning model 16 created or updated by the learning process shown in FIG. 7 Te, estimates the epicentral distance and focal depth (step B14).
[0090]
　The execution of steps B11 ~ B14, also in Embodiment 2, as in the first embodiment, based on the waveform data acquired from a single seismic detector 20, estimated epicentral distance and focal depth It is is will be.
[0091]
Effects of Second Embodiment
　As described above, in the second embodiment, the pretreatment with the waveform pre-processing section 41, the waveform data used for learning, suppression of the noise, the manifestation of features takes place . Therefore, according to the second embodiment improves the accuracy of the learning model, results would be achieved even improve the estimation accuracy.
[0092]
(Physical configuration)
　Here, by executing the program in the first and second embodiments, the computer realizing the epicenter distance estimation apparatus will be described with reference to FIG. Figure 9 is a block diagram showing an example of a computer that realizes the epicenter distance estimation apparatus according to Embodiment 1 and 2 of the present invention.
[0093]
　As shown in FIG. 9, the computer 110 includes a CPU 111, a main memory 112, a storage device 113, an input interface 114, a display controller 115, a data reader / writer 116, and a communication interface 117. These units, via a bus 121, are connected to each other to enable data communication.
[0094]
　CPU111 is stored in the storage device 113, a program of the present embodiment (the code) is expanded in the main memory 112 by executing them in a predetermined order to perform the various operations. The main memory 112 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory). The program in this embodiment is provided in a state stored in a computer readable recording medium 120. The program in this embodiment may be those flows on connected via the communication interface 117 the Internet.
[0095]
　Specific examples of the storage device 113, other hard disk drives, and semiconductor memory devices such as flash memory. Input interface 114 mediates the CPU 111, the data transmission between the input device 118 such as a keyboard and mouse. Display controller 115 is connected to a display device 119, controls display on the display device 119.
[0096]
　Data reader / writer 116 mediates the data transmission between the CPU111 and the recording medium 120, readout of the program from the recording medium 120, and executes the writing to the recording medium 120 of the processing result in the computer 110. Communication interface 117 includes a CPU 111, which mediates data transmission between the other computers.
[0097]
　Specific examples of the recording medium 120, CF (Compact Flash (registered trademark)) and SD (Secure Digital) general semiconductor recording device, a flexible disk (Flexible Disk) magnetic recording medium such as, or CD- the optical recording medium such as a ROM (Compact Disk Read Only Memory) and the like.
[0098]
　Further, epicentral distance estimation apparatus 10 and 40 in the first and second embodiments is not a computer program is installed, it can be realized by using the hardware corresponding to the respective parts. Furthermore, epicentral distance estimation apparatus 10 and 40, part of which is realized by a program, the remaining part may be implemented in hardware.
[0099]
　Some or all of the above-described embodiment are described below (Note 1) can be represented by - (Supplementary Note 15), but is not limited to the following description.
[0100]
(Supplementary Note 1)
　to obtain the waveform data of the generated seismic, and earthquake information acquisition unit,
　is applied to a learning model obtained by learning the relationship between waveform data and epicentral distance earthquake, the acquired the waveform data has been Te, estimates the epicenter distance, and estimation processing unit,
and a, epicentral distance estimation apparatus characterized by.
[0101]
(Supplementary Note 2)
　as input data the waveform data of the earthquake epicenter distance of the earthquake as the correct data, to learn the relationship between the waveform data and the epicentral distance, generates a learning model that shows a learning result, further comprising a learning unit ,
　the estimation processing unit, the learning model generated by said learning section, by applying the acquired the waveform data, to estimate the epicenter distance,
epicentral distance estimation apparatus according to Appendix 1.
[0102]
(Supplementary Note 3)
　The learning unit, in addition to the waveform data, the even location data of the point where the waveform data is obtained as input data, to learn the relationship between the epicenter distance between the waveform data and the location data the generated learning model,
　the earthquake information acquisition unit, in addition to the waveform data, also obtains location data of the point where the waveform data of the generated seismic was obtained,
　the estimation processing unit, the learning model against it, in addition to the waveform data, the location data may be applied obtained, it estimates the epicenter distance,
epicentral distance estimation apparatus according to note 2.
[0103]
(Supplementary Note 4)
　The learning processing unit, as the correct answer data, the focal depth further using earthquake, by learning the relationship between the depth of the epicenter distance and the epicenter and the waveform data, said learning model generates,
　the estimation processing unit, in addition to the epicenter distance, focal depth is also estimated,
epicentral distance estimation apparatus according to note 2 or 3.
[0104]
(Supplementary Note 5)
　The learning processing section, by learning to build a neural network, wherein the learning model of the neural network,
epicentral distance estimation apparatus according to any one of Appendixes 2-4.
[0105]
(Supplementary Note 6)
　the learning unit, for each waveform of the waveform data to be input data, generates the learning model,
　the estimation processing unit calculates a waveform of acquired the waveform data, the calculated waveform based on the amount, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, estimates the epicenter distance,
epicentral distance estimation apparatus according to any one of Appendixes 2-5.
[0106]
(Supplementary Note 7)
　The learning unit for each observation point of the waveform data to be input data, said generating a learning model,
　the estimation processing unit, to identify observation points acquired the waveform data, observations identified based on the point, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, estimates the epicenter distance,
epicentral distance estimation apparatus according to any one of Appendixes 2-5.
[0107]
(Supplementary Note 8)
　The learning unit for each soil characteristics at the observation point of the waveform data to be input data, said generating a learning model,
　the estimation processing unit, the ground characteristics of observation points acquired the waveform data has been identified, based on the specified ground characteristics, from each generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, the estimating the epicenter distance,
epicentral distance estimation apparatus according to any one of Appendixes 2-5.
[0108]
(Supplementary Note 9)
　waveform data is used wherein as the input data in said learning section, and the waveform data acquired in the seismic information acquisition unit, perform preprocessing further comprises a waveform pre-processor,
Appendix epicentral distance estimation apparatus according to any one of 1-8.
[0109]
(Supplementary Note 10)
　The waveform preprocessing unit, as the pretreatment, the image conversion process, an envelope conversion process, the bandpass conversion process, a differential conversion, and of the Fourier transformation process, performs at least one process,
Appendix 9 epicentral distance estimation apparatus according to.
[0110]
(Supplementary Note 11)
(a) acquiring the waveform data of the generated seismic, steps and,
(b) a learning model obtained by learning the relationship between waveform data and epicenter distance earthquake, the acquired waveform data by applying, to estimate the epicenter distance, the steps
having, epicentral distance estimation method characterized by.
[0111]
(Supplementary Note 12)
(c) as input data the waveform data of the earthquake epicenter distance of the earthquake as the correct data, to learn the relationship between the waveform data and the epicentral distance, generates a learning model that shows a learning result, a step further comprising,
　at said step of (b), a learning model generated by said step of (c), by applying the obtained the waveform data, to estimate the epicenter distance,
epicenter distance of statement 11 estimation method.
[0112]
(Supplementary Note 13)
　In the step of the (c), in addition to the waveform data, as location data is also input data point in which the waveform data is obtained, the relationship between the epicenter distance between the waveform data and the location data learning, the generated training model,
　in said step of (a), in addition to the waveform data, also obtains location data of the point where the waveform data of the generated seismic is obtained,
　said step of (b) in, with respect to the training model, in addition to said waveform data, and applies acquired the location data, estimates the epicenter distance,
epicentral distance estimation method of statement 12.
[0113]
(Supplementary Note 14)
　In the step of the (c), as the correct answer data, the focal depth further using earthquake, by learning the relationship between the depth of the epicenter distance and the epicenter and the waveform data, the generates a learning model,
　at said step of (b), in addition to the epicenter distance, the focal depth is also estimated,
epicentral distance estimation method of statement 12 or 13.
[0114]
(Supplementary Note 15)
　In the step of the (c), by learning to build a neural network, and learning model the neural network,
epicentral distance estimation method according to any one of Appendixes 12-14.
[0115]
(Supplementary Note 16)
　In the step of the (c), for each waveform of the waveform data to be input data, the generated training model,
　calculated in the step of said (b), the waveform of the acquired the waveform data has been and, based on the calculated waveform amount, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, the epicenter distance estimates the,
epicentral distance estimation method according to any one of Appendixes 12-15.
[0116]
(Supplementary Note 17)
　In the step of the (c), for each observation point of the waveform data to be input data, generates the learning models,
　in said step of (b), specifying the observation point of the acquired waveform data and, based on the specified observation point, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, the epicenter distance estimates the,
epicentral distance estimation method according to any one of Appendixes 12-15.
[0117]
(Supplementary Note 18)
　In the step of the (c), for each soil characteristics at the observation point of the waveform data to be input data, generates the learning models,
　in said step of (b), the observation of the obtained the waveform data has been identify soil properties of points, based on the specified ground characteristics, from each generated the learning model, selecting the learning model used, the learning model selected, the acquired the waveform data has been applied to, estimates the epicenter distance,
epicentral distance estimation method according to any one of Appendixes 12-15.
[0118]
(Supplementary Note 19)
(d) waveform data to be used as the input data in said step of (c), and the waveform data acquired in said step of (a), executes the pre-processing further comprises a step ,
epicentral distance estimation method according to any one of Appendixes 11-18.
[0119]
(Supplementary Note 20)
　In the step of the (d), as the pretreatment, the image conversion process, an envelope conversion process, the bandpass conversion process, a differential conversion, and of the Fourier transformation process, performs at least one process,
Appendix epicentral distance estimation method described in 19.
[0120]
(Supplementary Note 21)
to a computer,
obtains the waveform data of the earthquake that (a) generating a step,
a learning model obtained by learning the relation between (b) waveform data and epicentral distance earthquakes were acquired the waveform data by applying, to estimate the epicenter distance, and step,
thereby to execute includes instructions, computer-readable recording medium a program.
[0121]
(Supplementary Note 22)
to the computer,
as input data the waveform data of (c) an earthquake, the epicenter distance of the earthquake as the correct data, to learn the relationship between the waveform data and the epicenter distance, generates a learning model indicating a learning result to step further execute,
　in the step of said (b), the learning model generated by step (c), by applying the obtained the waveform data, to estimate the epicenter distance,
Clause 21 computer readable medium according.
[0122]
(Supplementary Note 23)
　In the step of the (c), in addition to the waveform data, as location data is also input data point in which the waveform data is obtained, the relationship between the epicenter distance between the waveform data and the location data learning, the generated training model,
　in said step of (a), in addition to the waveform data, also obtains location data of the point where the waveform data of the generated seismic is obtained,
　said step of (b) in, with respect to the training model, in addition to said waveform data, and applies acquired the location data, estimates the epicenter distance,
a computer-readable recording medium according to Appendix 22.
[0123]
(Supplementary Note 24)
　In the step of the (c), as the correct answer data, the focal depth further using earthquake, by learning the relationship between the depth of the epicenter distance and the epicenter and the waveform data, the generates a learning model,
　at said step of (b), in addition to the epicenter distance, the focal depth is also estimated,
a computer-readable recording medium according to Appendix 22 or 23.
[0124]
(Supplementary Note 25)
　In the step of the (c), by learning to build a neural network, and learning model the neural network,
a computer-readable recording medium according to any one of Appendixes 22-24.
[0125]
(Supplementary Note 26)
　In the step of the (c), for each waveform of the waveform data to be input data, the generated training model,
　calculated in the step of said (b), the waveform of the acquired the waveform data has been and, based on the calculated waveform amount, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, the epicenter distance to estimate,
computer-readable recording medium according to any one of Appendixes 22-25.
[0126]
(Supplementary Note 27)
　In the step of the (c), for each observation point of the waveform data to be input data, generates the learning models,
　in said step of (b), specifying the observation point of the acquired waveform data and, based on the specified observation point, from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, the epicenter distance to estimate,
computer-readable recording medium according to any one of Appendixes 22-25.
[0127]
(Supplementary Note 28)
　In the step of the (c), for each soil characteristics at the observation point of the waveform data to be input data, generates the learning models,
　in said step of (b), the observation of the obtained the waveform data has been identify soil properties of points, based on the specified ground characteristics, from each generated the learning model, selecting the learning model used, the learning model selected, the acquired the waveform data has been applied to, estimates the epicenter distance,
a computer-readable recording medium according to any one of Appendixes 22-25.
[0128]
(Supplementary Note 29)
to the computer,
executes the waveform data to be used as input data, and the waveform data acquired in said step of (a), the pre-processing in step; (d) (c), step further execute,
computer-readable recording medium according to any one of Appendixes 21-28.
[0129]
(Supplementary Note 30)
　In the step of the (d), as the pretreatment, the image conversion process, an envelope conversion process, the bandpass conversion process, a differential conversion, and of the Fourier transformation process, performs at least one process,
Appendix computer readable medium according to 29.
[0130]
　Although the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiment. Configuration and details of the present invention, it is possible to make various modifications that those skilled in the art can understand within the scope of the present invention.
[0131]
　This application claims priority based on Japanese Patent Application No. 2016-136310, filed on July 8, 2016, the entire disclosure of which is incorporated herein.
Industrial Applicability
[0132]
　As described above, according to the present invention, it is possible to calculate the epicenter distance stably, and it is possible to shorten the calculation time. The present invention, the present invention is, in the event of an earthquake, as soon as possible, it is useful to a system where there is a need to distribute information about the earthquake.
DESCRIPTION OF SYMBOLS
[0133]
　10 epicentral distance estimation apparatus (Embodiment
　1) 11 earthquake information acquiring unit
　12 estimating unit
　13 the learning information acquisition unit
　14 learning unit
　15 storage unit
　16 learning model
　20 earthquake detection device
　30 seismic activities overall monitoring system
　40 epicentral distance estimation apparatus (embodiment
　2) 41 waveform preprocessing section
　110 computer
　111 CPU
　112 main memory
　113 storage device
　114 the input interface
　115 display controller
　116 data reader / writer
　117 communication interface
　118 input device
　119 display device
　120 recording medium
　121 bus

WE CLAIM

It acquires waveform data of the generated seismic, and earthquake information acquisition unit,
　the learning model obtained by learning the relationship between waveform data and epicentral distance earthquakes, by applying acquired the waveform data, epicenter distance to estimate the estimation processing unit,
and a, epicentral distance estimation apparatus characterized by.
[Requested item 2]
　The waveform data of earthquakes as input data, the epicenter distance of the earthquake as the correct data, to learn the relationship between the waveform data and the epicentral distance, generates a learning model that shows a learning result, further comprising a learning unit,
　the estimation process parts, the learning model generated by the learning unit, by applying the acquired the waveform data, to estimate the epicenter distance,
epicentral distance estimation apparatus according to claim 1.
[Requested item 3]
　The learning section, in addition to the waveform data, location data of the point where the waveform data is obtained as an input data, to learn the relationship between the epicenter distance between the waveform data and the location data, said learning model generates,
　the earthquake information acquisition unit, in addition to the waveform data, also obtains location data of the point where the waveform data of the generated seismic was obtained,
　the estimation processing unit, to the learning model, said in addition to the waveform data, and applies acquired the location data, estimates the epicenter distance,
epicentral distance estimation apparatus according to claim 2.
[Requested item 4]
　The learning processing unit, as the correct answer data using the focal depth of the earthquake further learns the relationship between the depth of the epicenter distance and the epicenter and the waveform data to generate the learning model,
　the estimation processing unit, in addition to the epicenter distance, focal depth is also estimated,
epicentral distance estimation apparatus according to claim 2 or 3.
[Requested item 5]
　The learning processing section, by learning to build a neural network, wherein the learning model of the neural network,
epicentral distance estimation apparatus according to any one of claims 2-4.
[Requested item 6]
　The learning portion, each waveform of the waveform data to be input data, said generating a learning model,
　the estimation processing unit calculates a waveform of obtaining said waveform data, based on the calculated waveform weight , from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, estimates the epicenter distance,
claims 2 to epicentral distance estimation apparatus according to any one of 5.
[Requested item 7]
　The learning portion, each observation point of the waveform data to be input data, generates the learning model,
　the estimation processing unit, to identify observation points acquired the waveform data, based on the specified observation point , from among the respective generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, estimates the epicenter distance,
claims 2 to epicentral distance estimation apparatus according to any one of 5.
[Requested item 8]
　The learning portion, each soil characteristics at the observation point of the waveform data to be input data, generates the learning model,
　the estimation processing unit identifies the soil characteristics of the observation point of obtaining said waveform data, identified based on the soil properties, from each generated the learning model, selecting the learning model used, the learning model selected by applying the obtained the waveform data, estimates the epicenter distance to,
epicentral distance estimation apparatus according to any one of claims 2-5.
[Requested item 9]
　The waveform data is used the as input data in the learning unit, and the waveform data acquired in the seismic information acquisition unit, perform preprocessing further comprises a waveform pre-processor,
according to claim 1 to 8 epicentral distance estimation apparatus according to any one of.
[Requested item 10]
　The waveform pre-processing unit, as the pretreatment, the image conversion process, an envelope conversion process, the bandpass conversion process, a differential conversion, and of the Fourier transformation process, performs at least one process,
according to claim 9 epicentral distance estimation apparatus.
[Requested item 11]
(A) acquiring the waveform data of the generated seismic, a step,
a learning model obtained by learning the relation between (b) waveform data and epicentral distance earthquakes, by applying the obtained the waveform data estimates the epicenter distance, the steps
having, epicentral distance estimation method characterized by.
[Requested item 12]
The computer,
to obtain the waveform data of the earthquake (a), the steps,
the waveform data obtained learning model, obtained by learning the relationship between waveform data and epicenter distance (b) an earthquake apply and estimates the epicenter distance, the steps,
including instructions to execute the computer-readable recording medium recording a program.