KR20190080712A - Methods for differentiation of earthquake signal and prediction of earthquake intensity using randomly generated artificial seismic training data for an arbitrary zone - Google Patents

Abstract

The present invention relates to an earthquake determination method using randomly generated seismic learning data for arbitrary seismic zones. The earthquake determination method, an earthquake determination method using an artificial neural network technique, performed by an earthquake intensity prediction system for predicting an earthquake intensity using seismic waves in a target zone, comprises the steps of: collecting past earthquake data of the target zone and deriving seismic wave characteristic data from the collected past earthquake data; performing a pseudo earthquake probability model which reflects the derived seismic wave characteristic data to randomly generate learning data on general vibration and seismic vibration for the target zone; providing the learning data generated through the pseudo earthquake probability model to artificial neural network learning; inputting, if an actual P wave is detected, the P wave to an artificial neural network to determine whether an earthquake occurs; and generating an artificial S wave to estimate a structure′s response through the artificial neural network technique and predicting a damage scale of the structure using the generated artificial S wave to provide an earthquake warning. Therefore, the earthquake determination method of the present invention can be applied to smart seismometers in a house, a building, or the like, to improve the accuracy and reliability of earthquake warnings. The earthquake determination method can be also employed in a structural health monitoring (SHM) technology in construction and civil engineering fields to analyze a seismic impact on a structure. Moreover, the earthquake determination method is applied to a construction site to enable quick early evacuation and facility control in case of a natural disaster on the construction site.

Description

TECHNICAL FIELD [0001] The present invention relates to a method for determining an earthquake using randomly generated seismic learning data and a system for predicting the earthquake using the generated random earthquake learning data, and a system for predicting the earthquake using the randomly generated earthquake learning data.

The present invention relates to an earthquake discrimination method using randomly generated seismic learning data for an arbitrary seismic zone and a system for predicting the earthquake using the seismic learning data. More particularly, the present invention relates to an artificial neural network The present invention relates to an earthquake discrimination method using randomly generated seismic learning data for an arbitrary seismic zone object capable of predicting the earthquake power of an earthquake,

In general, seismic design for major structures is being applied as a preemptive preventive measure to minimize earthquake damage of structures.

However, earthquake damage may occur due to the occurrence of ground vibration exceeding the safety standard in the vicinity of the earthquake due to the nature of the earthquake.

On the other hand, in order to recover quickly in case of earthquake damage, appropriate follow-up measures should be performed according to the quantitative estimation of the earthquake damage of the structure and degree of damage.

However, there is not much model and methodology for quantitatively estimating the extent of earthquake damage of structures in case of mid - scale earthquake.

Generally known methods for earthquake damages of structures are earthquake damage (damage) magnitude by ground motion size using the seismic vulnerability curve of structures derived from empirical, experimental, or analytical methods assuming only large earthquakes It is a method of estimation.

However, this method estimates the magnitude of earthquake damage by assuming only large - scale earthquakes. Therefore, there is a problem that the magnitude of the earthquake damage can be overestimated in a region where a small - scale earthquake is expected.

In addition, in the past, earthquake damage to local structures was predicted by using artificial neural network technique based on past earthquake data, which is mostly limited. However, in the case of a region where there is insufficient earthquake data in the past, There is a problem that the accuracy and reliability of the earthquake warning for the area are inevitably lowered.

The applicant of the present invention has developed a method of predicting seismic information using a new artificial neural network technique which can solve the problems of the related art as described above.

Korean Patent No. 10-1598417 " Method of system identification (SI) for predicting nonlinear behavior of a building structure " Korean Patent No. 10-1713696 entitled " Prediction Information System for Structural Behavior by Seismic Wave "

The present invention randomly generates learning data of P-wave earthquake vibration and general vibration using past earthquake data by region and uses learning data as input data of artificial neural network to determine seismic discrimination, characteristics of S wave, Provided is a method for judging whether or not an earthquake has occurred using randomly generated earthquake learning data and a system for predicting the earthquake damage.

The earthquake discrimination method using randomly generated earthquake learning data for an arbitrary earthquake zone object according to a preferred embodiment of the present invention includes an artificial neural network technique performed by an earthquake discrimination and an earthquake prediction system for predicting the earthquake using a seismic wave of a target area The method comprising: collecting past earthquake data of a target area and deriving seismic characteristic data from the collected past earthquake data; Performing a pseudo vibration probability model for randomly generating learning data of general vibration and seismic vibration for the target area by reflecting the derived seismic characteristic data; Providing learning data generated through the pseudo-vibration probability model to artificial neural network learning; Inputting the detected P wave into the artificial neural network to determine whether or not an earthquake has occurred; And generating an artificial S wave for estimating a structure response through the artificial neural network technique and estimating a damage scale of the structure using the artificial S wave to provide an earthquake warning.

), 에너지 분포(

), 초기 지진파 증폭 특성, 지진 성분 대표 진폭비를 포함하는 것을 특징으로 한다.According to another embodiment of the present invention, the characteristic data of the seismic wave includes an effective value for the P wave

), Energy distribution (

), An initial seismic amplification characteristic, and an earthquake component representative amplitude ratio.

According to another embodiment of the present invention, the pseudo earthquake probability model uses the energy distribution, the effective value, the amplitude increase ratio, and the amplitude ratio of the P wave as input data and uses representative values including Norm, Max and Min And the learning data of the general vibration and the earthquake vibration is generated through the dimension reduction process through the normalization of the input data and the principal component analysis.

)와 주파수 에너지 분포(

)를 가지는 지진진동의 학습 데이터를 생성하는 것을 특징으로 한다.According to another embodiment of the present invention, the pseudo-vibration probability model randomly sets frequency-dependent energy intensity conversion coefficients (Ra, Rbk) in the past seismic data, and uses the set Ra and Rbk to calculate seismic- Reflected RMS value

) And the frequency energy distribution (

And generating the learning data of the seismic vibrations.

According to another embodiment of the present invention, the pseudo-vibration probability model is characterized in that learning data is generated by applying a triangular window function for randomly adjusting an amplification ratio of an initial seismic wave to reflect an amplification characteristic of an initial seismic wave .

Further, the learning data is generated in consideration of the representative amplitude ratio of the horizontal and vertical components.

According to another embodiment of the present invention, the pseudo vibration probability model may include: outputting a first earthquake sample value in the frequency domain through Fast Fourier Transform (FFT) of the past earthquake data; Setting a random Ra and Rbk having a range (l1, l2) as a probability statistic characteristic of the past earthquake data to the first earthquake sample value, and then performing a frequency-to-frequency energy distribution conversion to output a second earthquake sample value; And converting the second seismic sample value into a time domain in the frequency domain by an inverse FFT to generate learning data of the seismic vibration.

)와 주파수 에너지 분포(

)를 가지는 일반진동의 학습 데이터를 생성하는 것을 특징으로 한다.According to another embodiment of the present invention, the pseudo-vibration probability model randomly sets frequency-dependent energy intensity conversion coefficients (Ra, Rbk) in the past seismic data, and uses the set Ra and Rbk to calculate seismic- Reflected RMS value

) And the frequency energy distribution (

And generating the learning data of the general vibration having the normal vibration.

According to another embodiment of the present invention, the pseudo vibration probability model may include a step of outputting a second general sample value in the frequency domain through time domain random scaling and Fast Fourier Transform (FFT) ; Setting a random Ra, Rbk set to a range (l1, l2) deviating from a probability statistic characteristic of the past seismic data to the first general sample value, and then performing a frequency energy distribution conversion to output a second general sample value; And converting the second normal sample value into a time domain in the frequency domain by inverse FFT to generate learning data of the general oscillation.

)를 계산하여 P파를 이용한 지진 판별 인공 신경망을 생성하는 단계; 및 P파의 학습 데이터를 이용하여 에너지 강도 및 분포에 대한 주파수 특성을 이용하여 S파의 주파수 특성을 인공 신경망으로 추정하여 S파 생성 인자 도출 인공 신경망을 생성하는 단계를 포함하는 것을 특징으로 한다.According to another embodiment of the present invention, the step of providing the learning data generated through the pseudo-vibration probability model to the artificial neural network learning includes a step of, when the learning data of the earthquake vibration and the general vibration using the P- Earthquake discrimination artificial neural network weighting

) To generate an earthquake discrimination artificial neural network using the P wave; And generating S-wave generating factor-derived artificial neural network by estimating the frequency characteristic of the S-wave using the artificial neural network using the frequency characteristics of the energy intensity and the distribution using the learning data of the P-wave.

,

이며,

의 정규화 실시, 입력-출력 데이터 집합에 의한 가중치(

)를 계산하는 것을 특징으로 한다.According to another embodiment of the present invention, the output value of the earthquake discrimination artificial neural network using the P wave is expressed by the following Equation 1,

,

Lt;

, Weighting by input-output data set (

) Is calculated.

,

이고,

의 정규화 실시 및 입력-출력 데이터 집합에 의한 가중치(

)를 계산하는 것을 특징으로 한다.According to another embodiment of the present invention, the output value of the S-wave generating factor artificial neural network is represented by the following equation (2), where R'a and R'bk are intensity conversion coefficients by S- Es is the energy distribution by S-wave frequency,

,

ego,

And weighting by the input-output data set (

) Is calculated.

According to another embodiment of the present invention, the step of generating an artificial S wave for estimating the structure response through the artificial neural network technique and estimating the damage scale of the structure using the artificial S wave to provide an earthquake alert , And an artificial S wave is generated using Equation (3).

According to another embodiment of the present invention, the step of generating an artificial S wave for estimating the structure response through the artificial neural network technique and estimating the damage scale of the structure using the artificial S wave to provide an earthquake alert , A maximum displacement and an inter-story displacement are derived by analyzing a non-linear structure using the artificial S wave, and an alarm level is set through the derived maximum displacement and inter-story displacement to generate an earthquake warning.

According to another embodiment of the present invention, the step of generating an artificial S wave for estimating the structure response through the artificial neural network technique and estimating the damage scale of the structure using the artificial S wave to provide an earthquake alert Generating structural response learning data by performing nonlinear analysis of a structure in advance using learning data of the general vibration and seismic vibration; And deriving a maximum displacement and an inter-story displacement using the frequency characteristics of the artificial S-wave and setting an alarm level through the derived maximum displacement and inter-story displacement to generate an earthquake warning.

According to another embodiment of the present invention, the maximum acceleration and the progress information are calculated using the last measured seismic data of the P wave and the S wave having been provided with the earthquake warning and stored in the database, And re-learning the neural network.

According to another embodiment of the present invention, the past earthquake data and the damage scale data by the earthquake scale are stored in the database, and the currently measured earthquake data is compared with the past earthquake data and the damage scale data by the earthquake scale, And estimating the magnitude of the earthquake and the magnitude of the damage caused by the earthquake.

According to another embodiment of the present invention, when the building response sensor is installed in the facility, the method further includes learning the damage scale of the facility due to the earthquake using the sensor data of the building response sensor and the last measured earthquake data .

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the term "comprises" or "having ", etc. is intended to specify the presence of stated features, integers, steps, operations, elements, parts, or combinations thereof, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

 According to the earthquake discrimination method and the earthquake prediction system using the random earthquake learning data of the arbitrary earthquake zone target of the present invention, it is possible to improve the accuracy and reliability of the earthquake warning applied to the smart seismometer of the facility, (SHM) technology to analyze the seismic effect of the structure, and it can be applied to the structure under construction, so that it is possible to perform the early evacuation and early facility control in case of natural disaster at the construction site.

FIG. 1 is a diagram illustrating a progress prediction system according to an embodiment of the present invention. FIG.
FIG. 2 is a flowchart illustrating an earthquake determination method using an artificial neural network technique according to an embodiment of the present invention. FIG.
3 is a flowchart illustrating an artificial neural network learning process using the past earthquake data of FIG. 2,
FIG. 4 is a flowchart for explaining a response derivation process of the seismic determination structure of FIG. 2;
5 is a view for explaining an earthquake scale and a damage scale prediction process using an earthquake discrimination method using randomly generated seismic learning data for an arbitrary seismic zone object according to an embodiment of the present invention.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a view for explaining a progress prediction system according to an embodiment of the present invention.

Referring to FIG. 1, the progress prediction system 100 includes an earthquake measurement unit 110, an earthquake analysis unit 120, an earthquake alert generator 130, and a database 140.

The earthquake measuring unit 110 senses the general vibration as well as the P wave and the S wave and provides it to the seismic analysis unit 120.

The seismic analysis unit 120 uses an artificial neural network technique to predict the magnitude of the S wave and the progress of the structure in the P wave measurement, reflecting the characteristics of the target area using the past earthquake data.

The seismic analysis unit 120 performs a pseudo vibration probability model for randomly generating learning data of the general vibration and the seismic vibration for the target area by reflecting the seismic characteristic data of the target area, Wave, the magnitude of real-time S-wave, the response level of the structure, and the progress level of the stratum.

In this case, the Pseudo Earthquake probability model uses the energy distribution, effective value, amplitude increase ratio, and amplitude ratio of the P wave as input data, and normalization and principal component analysis of input data using representative values such as Norm, Max, Is a model that generates learning data of general vibration and seismic vibration through the dimensional reduction process.

The seismic alarm generating unit 130 predicts the damage scale of the structure analyzed by the seismic analysis unit 120 and generates an earthquake warning including an early evacuation alarm or early facility control.

The database 140 stores past earthquake data, damage scale data for the past earthquake scale, and currently measured earthquake data for each region.

2 is a flowchart illustrating an earthquake discrimination method using an artificial neural network technique according to an embodiment of the present invention.

2, the earthquake discrimination method using the artificial neural network technique collects past earthquake data of the target area and derives seismic characteristic data from the collected past earthquake data (S201)

)의 특성 데이터는 P파에 대한 실효치(

), 에너지 분포(

), 초기 지진파 증폭 특성, 지진 성분 대표 진폭비를 포함한다. 여기서,

는 i번째 지진파의 FFT 결과에서 k번째 주파수 그룹을 의미한다.i-th past P-wave earthquake record data

) Is the rms value of the P wave (

), Energy distribution (

), Initial seismic amplification characteristics, and seismic component representative amplitude ratio. here,

Denotes the k-th frequency group in the FFT result of the i-th seismic wave.

 In this case, the root mean square (RMS) is a value that is a basic value when probabilistic statistical evaluation of arbitrary data, that is, averaging time, Can be expressed. Note that the rms value varies depending on the averaging time.

The pseudo-prediction system 100 performs a pseudo-vibration probability model that randomly generates learning data of the general vibration and the seismic vibration for the target area by reflecting the derived seismic characteristic data, To the artificial neural network learning (S202, S203)

When the actual P-wave is detected, the system predicts the earthquake by inputting the P-wave earthquake signal to the learned artificial neural network (S204, S205, S206)

In addition, the progress prediction system generates an artificial S wave for estimating the structure response by deriving the characteristics of the S wave through the learned artificial neural network (S207)

The prediction system estimates the damage level of the structure using the artificial S wave and issues an earthquake warning (S208, S209)

3 is a flowchart illustrating an artificial neural network learning process using the past earthquake data of FIG.

Referring to FIG. 3, in the case of the region where the medium-sized earthquake data is insufficient, the pseudo-prediction system uses the past earthquake data for the adjacent region as the initial learning data, And generates learning data of the general vibration (S301, S302)

)와 주파수 에너지 분포(

)를 가지는 지진진동의 학습 데이터를 생성한다.At this time, the pseudo-vibration probability model randomly sets frequency-dependent energy intensity conversion coefficients (Ra, Rbk) in the past earthquake data, and uses the set Ra and Rbk to calculate the rms value

) And the frequency energy distribution (

) Of the earthquake vibration.

,

,

)를 생성한다.That is, the pseudo vibration probability model outputs the first earthquake sample value in the frequency domain through the Fast Fourier Transform (FFT) of the past earthquake data, and outputs the first earthquake sample value as the probability statistic characteristic of the past earthquake data After setting the random Ra and Rbk in the range (11, 12), the frequency energy distribution transformation is performed to output the second seismic sample value, and the second seismic sample value is transformed from the frequency domain to the time domain by the inverse FFT Learning data of earthquake vibration (

,

,

).

)와 주파수 에너지 분포(

)를 가지는 일반진동의 학습 데이터를 생성한다.On the other hand, the pseudo-vibration probability model randomly sets frequency-dependent energy intensity conversion coefficients (Ra, Rbk) in the past earthquake data and uses Ra and Rbk to calculate the rms value

) And the frequency energy distribution (

) Of the normal vibration.

의 초기 임의 시간대에 삼각형 윈도우(window) 함수를 적용하여 변화시킬 수 있으며, 상기 삼각형 윈도우 함수를 사용하여 학습 데이터를 생성한다.In detail, to reflect changes in the initial seismic intensity amplification ratio

A triangle window function may be applied at an initial arbitrary time zone of the triangle window function to generate learning data using the triangle window function.

At this time, the learning data can be generated in consideration of the representative amplitude ratio of the horizontal and vertical components.

,

,

)를 생성한다.That is, the pseudo vibration probability model outputs the first general sample value in the frequency domain through the time domain random scaling and the fast Fourier transform (FFT) of the past earthquake data, and adds the probability statistical characteristic And outputs a second normal sample value by performing a frequency energy distribution conversion after setting the random Ra and Rbk set to a range (l1, l2) out of the frequency domain to a time domain from the frequency domain by an inverse FFT And transforms the learning data (

,

,

).

Thereafter, the progress prediction system provides the learning data of the seismic vibration and the general vibration using the P wave generated through the pseudo-vibration probability model as input data of the artificial neural network, and performs the dimension reduction process (S303).

Also, the S / P system generates the S-wave generating factor artificial neural network by estimating the frequency characteristic of the S-wave using the artificial neural network using the frequency characteristics of the energy intensity and the distribution using the P-wave learning data (S304)

)를 계산하고, S파 생성 인자 도출 인공 신경망 가중치(

)를 계산한다.(S305, S306)Progression prediction system is based on earthquake discrimination artificial neural network weighting

), And the S-wave generation factor derived artificial neural network weight (

) (S305, S306)

 In this case, the output value of the earthquake discrimination artificial neural network using the P wave can be expressed by the following equation (1).

[Equation 1]

,

이며,

의 정규화 실시(MinMax, STD, norm), 입력-출력 데이터 집합에 의한 가중치(

)를 계산한다.In Equation (1)

,

Lt;

(MinMax, STD, norm), weight by input-output data set (

).

,

이고,

의 정규화 실시(MinMax, STD, norm) 및 입력-출력 데이터 집합에 의한 가중치(

)를 계산한다.Further, the output value of the S-wave generating factor-derived artificial neural network is expressed by the following equation (2)

,

ego,

Normalization (MinMax, STD, norm) and input-output data set weight (

).

Here, R'a and R'bk are the intensity conversion coefficients for the S-wave frequency, and Es is the energy distribution for the S-wave frequency.

&Quot; (2) "

4 is a flowchart for explaining a response derivation process of the seismic discrimination structure of FIG.

Referring to FIG. 4, the progress prediction system measures an earthquake signal by measuring a real-time P wave, provides a P-wave earthquake signal as an input data to the artificial neural network, (S401, S402, S403)

The earthquake prediction system determines earthquake through an output value for earthquake discrimination, calculates an output value for generating an artificial S wave in the artificial neural network when it is judged to be an earthquake, and uses an artificial S wave (S405, S406)

&Quot; (3) "

Since the characteristic relation between the P wave and the S wave has a nonlinear relationship with the ground, the fault layer characteristics, and the depth of occurrence of the object zone, the PPC system performs nonlinear analysis of the structure using the artificial S wave, And an earthquake warning for early evacuation alarm and early facility control are generated (S407, S408, S410) by calculating the level of progress (floor displacement) and the level of stratification (interlayer displacement)

을 입력 데이터로 제공하고, 인공 신경망에서 구조물의 최대 변위 및 층간 변위를 도출할 수도 있다. (S409)Or propagation prediction system generates the structure response learning data by performing the nonlinear analysis of the structure in advance using the learning data of the general vibration and the seismic vibration,

As the input data, and derive the maximum displacement and interstory displacement of the structure in the artificial neural network. (S409)

The system predicts the maximum acceleration and the progress information by using the last measured seismic data of the P wave and the S wave that have been provided with the seismic warning and stores the maximum acceleration and the progress information in the database 140. Then, (S411, S412)

At this time, not only the final measured seismic data but also the data related to the general vibration are used as the re-learning data, and it is reflected in the seismic characteristic data of the target area so as to continuously improve the accuracy of seismic discrimination and earthquake damage estimation.

5 is a view for explaining an earthquake scale and a damage scale prediction process using an earthquake discrimination method using randomly generated seismic learning data for an arbitrary seismic zone object according to an embodiment of the present invention.

5, the magnitude prediction system stores past earthquake data in a database for each region, and also stores damage scale data for each past earthquake scale in the database (S501, S502)

When the actual P-wave earthquake signal is measured, the magnitude prediction system reads past earthquake data, damage data by past earthquake scale, and compares it with the currently measured earthquake data to predict future earthquake magnitude and damage scale. S503, S504, S505)

Also, in the case where the building response sensor is installed in the facility, the progress prediction system can learn the scale of the damage caused by the earthquake using the sensor data of the building response sensor and the last measured earthquake data.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the present invention as defined by the following claims It can be understood that

It should be noted that the scope of the present invention is not limited to the embodiments described in the description of the present invention, And is not to be construed as limited by the drawings. That is, the embodiments are to be construed as being variously embodied and having various forms, so that the scope of the present invention should be understood to include equivalents capable of realizing technical ideas. Also, the purpose or effect of the present invention should not be construed as limiting the scope of the present invention, since it does not mean that a specific embodiment should include all or only such effect.

Further, all terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs, unless otherwise defined. Commonly used predefined terms should be interpreted to be consistent with the meanings in the context of the relevant art and can not be construed as having ideal or overly formal meanings which are not expressly defined in the present invention.

110: Seismic measurement part 120: Seismic analysis part
130: Seismic alert generator 140: Database

Claims (17)

In an earthquake discrimination method using an artificial neural network technique performed by an earthquake prediction system for predicting an earthquake using a seismic wave of a target area,
Collecting past earthquake data of a target area and deriving seismic characteristic data from the collected past earthquake data;
Performing a pseudo vibration probability model for randomly generating learning data of general vibration and seismic vibration for the target area by reflecting the derived seismic characteristic data;
Providing learning data generated through the pseudo-vibration probability model to artificial neural network learning;
Inputting the detected P wave into the artificial neural network to determine whether or not an earthquake has occurred; And
Generating an artificial S wave for estimating a structure response through the artificial neural network technique and estimating a damage scale of the structure using the generated artificial S wave to provide an earthquake warning; A Method for Determining Earthquake Using Random Generation Earthquake Learning Data.
The method according to claim 1,
The seismic wave (

) Is the rms value of the P wave (

), Energy distribution (

), An initial seismic amplification characteristic, and a representative amplitude ratio of an earthquake component.
3. The method of claim 2,
The pseudo earthquake probability model uses the energy distribution, the effective value, the amplitude increase ratio, and the amplitude ratio of the P wave as the input data, and normalization and principal component analysis of the input data using the representative values including Norm, Max and Min And generating learning data of the general vibration and the earthquake vibration through the dimension reduction process. The earthquake discrimination method using the random earthquake learning data for the arbitrary earthquake zone object.
3. The method of claim 2,
Wherein the pseudo-vibration probability model comprises:
Frequency-dependent energy intensity conversion coefficients (Ra, Rbk) are randomly set in the past seismic data, and an erroneous value

) And the frequency energy distribution (

And generating learning data of the earthquake vibration having the random seismic zone object randomly generated.
3. The method of claim 2,
Wherein the pseudo-vibration probability model comprises:
And a triangular window function for randomly adjusting the amplification ratio of the initial seismic wave to reflect the amplification characteristics of the initial seismic wave. Identification method.
5. The method of claim 4,
Wherein the pseudo-vibration probability model comprises:
Outputting a first earthquake sample value in the frequency domain through Fast Fourier Transform (FFT) of the past earthquake data;
Setting a random Ra, Rbk with a range (l1, l2) as a probability statistic characteristic of the past earthquake data to the first earthquake sample value, and then performing a frequency energy distribution transformation to output a second earthquake sample value; And
And generating learning data of an earthquake vibration by converting the second earthquake sample value into a time domain in a frequency domain by an inverse FFT to generate seismic vibration learning data. .
3. The method of claim 2,
Wherein the pseudo-vibration probability model comprises:
Frequency-dependent energy intensity conversion coefficients (Ra, Rbk) are randomly set in the past seismic data, and the effective values (Ra, Rbk)

) And the frequency energy distribution (

And generating learning data of the general vibration having the random vibration generating seismic learning data.
8. The method of claim 7,
Wherein the pseudo-vibration probability model comprises:
Outputting a second general sample value in the frequency domain through time domain random scaling and Fast Fourier Transform (FFT) of the past earthquake data;
Setting a random Ra, Rbk set to a range (l1, l2) deviating from a probability statistic characteristic of the past seismic data to the first general sample value, and then performing a frequency energy distribution conversion to output a second general sample value; And
And generating learning data of a general vibration by converting the second general sample value into a time domain in a frequency domain by an inverse FFT to generate a learning data of a general vibration. .
The method according to claim 1,
Wherein the step of providing the learning data generated through the pseudo-vibration probability model to the artificial neural network learning comprises:
If the learning data of seismic vibration and general vibration using P wave is input to artificial neural network, the earthquake discrimination artificial neural network weighting

) To generate an earthquake discrimination artificial neural network using the P wave; And
And generating an S-wave generating factor-derived artificial neural network by estimating the frequency characteristic of the S-wave using the artificial neural network using the frequency characteristics of the energy intensity and the distribution using the P-wave learning data. A Method for Determining Earthquake Using Random Generation Earthquake Learning Data.
10. The method of claim 9,
The output value of the earthquake discrimination artificial neural network using the P wave is expressed by the following equation (1)
[Equation 1]

In Equation (1)

,

Lt;

, Weighting by input-output data set (

) Is calculated based on the seismic learning data of the arbitrary seismic zone object.
10. The method of claim 9,
The output value of the S-wave generating factor derived artificial neural network is expressed by the following equation (2)
&Quot; (2) "

In Equation (2), R'a and R'bk are intensity conversion coefficients by S-wave frequency, Es is an energy distribution by S-wave frequency,

,

ego,

And weighting by the input-output data set (

) Is calculated based on the seismic learning data of the arbitrary seismic zone object.
The method according to claim 1,
Generating an artificial S wave for estimating a structure response through the artificial neural network technique and estimating a damage scale of the structure using the artificial S wave to provide an earthquake alert,
&Quot; (3) "

And generating an artificial S wave by using Equation (3).
The method according to claim 1,
Generating an artificial S wave for estimating a structure response through the artificial neural network technique and estimating a damage scale of the structure using the artificial S wave to provide an earthquake alert,
And a seismic alert is generated by setting an alarm level through the derived maximum displacement and the inter-story displacement. The method of claim 1, A Method for Determining Whether an Earthquake Occurred Using Generated Seismic Learning Data.
The method according to claim 1,
Generating an artificial S wave for estimating a structure response through the artificial neural network technique and estimating a damage scale of the structure using the artificial S wave to provide an earthquake alert,
Generating structure response learning data by performing nonlinear analysis of a structure in advance using learning data of the general vibration and seismic vibration; And
And deriving a maximum displacement and an inter-story displacement using the frequency characteristic of the artificial S wave and setting an alarm level through the derived maximum displacement and inter-story displacement to generate an earthquake warning A Method for Determining Whether an Earthquake Occurred Using Target Randomly Generated Seismic Learning Data.
The method according to claim 1,
Calculating the maximum acceleration and the progress information using the last measured seismic data of the P wave and the S wave having been provided with the earthquake warning and storing the calculated maximum acceleration and the progress information in a database and re-learning the artificial neural network using the last measured earthquake data And determining whether or not the earthquake has occurred by using randomly generated earthquake learning data.
16. The method of claim 15,
By storing historical earthquake data and damage data by seismic scale in each region and comparing the present earthquake data with past earthquake data and damage data by seismic scale, Further comprising the step of determining whether or not the seismic region is randomly generated.
16. The method of claim 15,
Further comprising the step of, when the building response sensor is installed in the facility, learning the damage scale of the facility due to the earthquake using the sensor data of the building response sensor and the last measured earthquake data. A Method for Determining Whether an Earthquake Occurred Using Generated Seismic Learning Data.

Images