KR20150035633A - Apparatus for measuring earthquake intensity and method for the same - Google Patents

Abstract

The present invention relates to an earthquake measuring apparatus and an earthquake measuring method. According to an embodiment of the present invention, the earthquake measuring apparatus comprises: a seismometer producing an earthquake signal by sensing a seismic wave produced by an earthquake; and an analyzing portion deducting the magnitude of the earthquake by calculating a characteristic parameter showing a feature of the earthquake from the earthquake signal, and by inputting the characteristic parameter in an artificial neutral net function.

Description

FIELD OF THE INVENTION [0001] The present invention relates to an earthquake measuring apparatus,

The present invention relates to an earthquake measuring apparatus and an earthquake measuring method, and more particularly, to an earthquake measuring apparatus and an earthquake measuring method capable of deriving the magnitude of an earthquake using an artificial intelligence network circuit.

If an earthquake occurs, industrial facilities and other structures may be damaged, and damage to people may be accompanied. At this time, the degree of damage caused by an earthquake can vary depending on the magnitude of the earthquake and the distance between the earthquake occurrence point (origin). In general, the energy of earthquakes is attenuated and weakened as the distance from the point of occurrence of earthquakes increases. Therefore, when an earthquake occurs, it is very important to measure the magnitude and distance of the earthquake as quickly as possible.

Generally, the magnitude of earthquake and the method of measuring the distance are calculated by first calculating the distance of the earthquake using the difference between the arrival time of the P wave arriving first among the seismic waves and the arrival time of the S wave arriving late, The magnitude of the earthquake is calculated using the damping equation of the magnitude of the seismic wave. In this case, in order to measure the difference between the arrival times of the P wave and the S wave, it is necessary to be able to distinguish between the P wave and the S wave. However, when the earthquake waveform is bad, This can be difficult.

In addition, there is a method of estimating an earthquake scale using the magnitude of the P wave and measuring the distance to the earthquake occurrence point by inputting the magnitude of the S wave in the magnitude-distance attenuation equation of the seismic wave. That is, it performs the function of earthquake early warning. However, in this case, since the magnitude of the P wave is used to predict the magnitude of the earthquake, an error of the magnitude of the earthquake scale may occur. Further, when the level of the noise signal included in the seismic waveform is high, it is difficult to distinguish it from the P wave.

Open Patent Publication No. 2012-0033382 (Apr. 04, 2012)

This application proposes an earthquake measuring apparatus and an earthquake measuring method capable of deriving the magnitude of an earthquake using an artificial intelligence network circuit.

A method of measuring an earthquake according to an embodiment of the present invention includes a seismic wave sensing step of sensing an earthquake wave generated by an earthquake to generate an earthquake signal; A characteristic parameter calculation step of calculating a characteristic parameter indicating a characteristic of the earthquake from the seismic signal; And inputting the characteristic parameter into an artificial neural network (ANN) function to derive an earthquake magnitude of the earthquake.

Here, the characteristic parameter calculation step may include calculating a characteristic parameter such as Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV), Cumulative Absolute Velocity (CAV), and Cumulative Absolute Velocity : Cumulative Absolute Displacement).

Here, the characteristic parameter calculation step may calculate the maximum value among the ground acceleration and the ground velocity measured during the earthquake duration, and set the maximum ground acceleration and the maximum ground velocity.

Wherein the characteristic parameter calculation step comprises:

,

,

, t는 시간, a_th는 기준지반가속도, v_th는 기준지반속도, N은 지진지속시간, n_s는 단위시간당 샘플링수를 이용하여, 상기 누적절대속도 및 누적절대변위를 계산할 수 있다. Where CAV is the cumulative absolute velocity, CAD is the cumulative absolute displacement,

,

,

, t is the time, a _th is the reference ground acceleration, v _th is the reference ground velocity, N is the seismic duration, and n _s is the number of samples per unit time.

Here, the earthquake scale derivation step may use a two layer feed-forward (TLFF) function as the artificial neural network function.

The earthquake scale derivation step includes an input layer including (P + 1) input nodes, a hidden layer including (Q + 1) intermediate nodes, and one output node (P + 1) * (Q + 1) first weights for each intermediate node associated with the input node and an output node associated with the intermediate node Q + 1) second weighting values, the characteristic parameters are associated with each of the input nodes, and the earthquake scale is associated with the output nodes, thereby forming an artificial neural network.

Here, the earthquake scale derivation step may include:

y is between the output node, x _p the p-th input node, w ⁽¹⁾ _qp is the first weight between the p-th input node and the q-th intermediate node, w ⁽²⁾ _q is the q-th intermediate node and the output node The earthquake magnitude can be derived using an artificial neural network function with a second weight, x ₀ = 1.

Here, the earthquake measuring method includes learning the artificial neural network function using characteristic parameters and an earthquake scale included in the measured earthquake observation data, and outputting the corresponding earthquake scale when the characteristic parameter is input to the artificial neural network function And setting the first weight and the second weight to set the first weight and the second weight.

Here, the earthquake scale derivation step may include:

를 상기 입력노드에 각각 입력하여 상기 지진규모를 도출할 수 있다. 여기서, 상기 Vs은 지표로부터 기 설정된 깊이까지의 평균전단파속도일 수 있다.

Can be input to the input node to derive the earthquake magnitude. Here, Vs may be an average shear wave velocity from the surface to a predetermined depth.

Here, the earthquake scale derivation step may include inputting a characteristic parameter, which is obtained by applying a plurality of different reference ground accelerations and reference ground speeds, to the artificial neural network function, calculating an earthquake magnitude corresponding to the plurality of reference ground accelerations and the reference ground speed, A first step of deriving a candidate group of the input signal; And a second step of inputting the candidate group to a second artificial neural network function to derive one earthquake scale value from among the earthquake magnitudes included in the candidate group.

The first step may include calculating an average value of the earthquake magnitude calculated when the accumulated absolute velocity is not 0 when the value of the accumulated absolute velocity is calculated as 0 according to the value of the reference ground acceleration, Can be set.

The first step may terminate the calculation of the earthquake scale when the value of the accumulated absolute displacement is calculated as 0 according to the value of the reference ground speed.

Here, the second artificial neural network function is a function of the seismic intensity of the earthquake scale closest to the actual earthquake magnitude stored in the seismic observation data among the plurality of seismic scale values according to the reference ground acceleration and the reference ground velocity, Can be learned.

An apparatus for measuring an earthquake according to an embodiment of the present invention includes: a seismometer for generating an earthquake signal by sensing a seismic wave generated by an earthquake; And an analyzer for calculating a characteristic parameter representing a characteristic of the earthquake from the seismic signal and inputting the characteristic parameter to an artificial neural network function to derive an earthquake magnitude of the earthquake.

Here, the analysis unit may include at least the maximum ground speed and the cumulative absolute displacement as the characteristic parameters.

Here, the analyzing unit inputs characteristic parameters, which are respectively applied with a plurality of different reference ground accelerations and reference ground speeds, to the artificial neural network function, and calculates a candidate group of earthquake magnitudes calculated corresponding to the plurality of reference ground acceleration and reference ground speed The candidate group is input to the second artificial neural network function, and one earthquake scale value among the earthquake magnitudes included in the candidate group can be derived.

In addition, the means for solving the above-mentioned problems are not all enumerating the features of the present invention. The various features of the present invention and the advantages and effects thereof will be more fully understood by reference to the following specific embodiments.

According to the earthquake measuring apparatus and the earthquake measuring method according to an embodiment of the present invention, since the earthquake scale is calculated using the ground acceleration sensitive to high frequency and the ground speed sensitive to low frequency, it is possible to measure the earthquake scale reflecting the broadband frequency characteristic , The reliability of the measured earthquake magnitude can be improved.

According to the earthquake measuring apparatus and the earthquake measuring method according to the embodiment of the present invention, since the earthquake scale is calculated using the ground speed, the effect due to the high frequency noise and the stress reduction amount is minimized, Distortion can be minimized.

According to the earthquake measuring apparatus and the earthquake measuring method according to the embodiment of the present invention, since the reference ground speed can be set according to the earthquake magnitude class, the earthquake magnitude can be selectively calculated for the earthquake more than the calamity earthquake.

1 is a block diagram showing an apparatus for measuring an earthquake according to an embodiment of the present invention.
2 is a graph showing characteristic parameters appearing in an earthquake signal according to an embodiment of the present invention.
3 is a diagram illustrating an artificial neural network according to an embodiment of the present invention.
4 is a schematic view showing an operation of an analysis unit of an earthquake measuring apparatus according to an embodiment of the present invention.
5 is a flowchart showing an earthquake measuring method according to an embodiment of the present invention.
6 to 8 are comparison charts and graphs showing seismic scale calculation results obtained by an earthquake measuring method according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, in order that those skilled in the art can easily carry out the present invention. In the following detailed description of the preferred embodiments of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. In the drawings, like reference numerals are used throughout the drawings.

In addition, in the entire specification, when a part is referred to as being 'connected' to another part, it may be referred to as 'indirectly connected' not only with 'directly connected' . Also, to "include" an element means that it may include other elements, rather than excluding other elements, unless specifically stated otherwise.

1 is a block diagram showing an apparatus for measuring an earthquake according to an embodiment of the present invention.

Referring to FIG. 1, an apparatus for measuring an earthquake according to an embodiment of the present invention may include a seismograph 10 and an analysis unit 20.

Hereinafter, an earthquake measuring apparatus according to an embodiment of the present invention will be described with reference to FIG.

The seismometer 10 can generate an earthquake signal by sensing a seismic wave generated by an earthquake. The seismic system can be applied to any commonly used seismic system. Specifically, the seismometer can measure seismic waves vibrating in a horizontal uniaxial direction, and can measure the magnitudes of the ground speed and the ground acceleration of the seismic waves from the measured seismic waves.

Herein, the seismometer includes both an earthquake velocity meter for measuring the ground speed and an earthquake accelerometer for measuring the ground acceleration, or integrating the measured ground acceleration with respect to time, It may be to obtain speed.

Generally, since the ground speed is a low-frequency signal, it is robust to a high-frequency noise (about 10 Hz or more) included in a seismic wave. Therefore, if the earthquake analysis is performed using the ground speed, the distortion of the earthquake analysis result according to the noise signal can be minimized. Also, since the signal-to-noise ratio of the P wave is good in the case of the ground speed, it is possible to perform the earthquake analysis using the P wave, so that the earthquake early warning can be performed before the S wave reaching the structure.

Performing the seismic analysis using both the ground speed and the ground acceleration means conceptually using the high frequency component seismic wave and the low frequency component seismic wave at the same time. Since a general seismic wave corresponds to a broadband frequency waveform having a low frequency and a high frequency component at the same time, it is possible to improve the reliability of the seismic analysis result by simultaneously using the ground speed and the ground acceleration.

Furthermore, by using the ground speed, it is possible to estimate the seismic scale stably by minimizing the distortion caused by the stress drop (Δσ). Specifically, the sensitivity of earthquake scale estimation to the maximum value (a _max ) of the ground acceleration and the stress drop (Δσ) of the maximum value of the ground velocity (v _max ) is as follows.

As shown in the above equation, since the log of the maximum value v _max of the ground speed is 0.8 times less sensitive to the log Δσ log than the log of the maximum a _max of the ground acceleration a _max , it is possible to minimize the seismic scale estimation distortion . On the other hand, the log (V _max ) is 1.8 times more sensitive than the log (a _max ) for the seismic scale, so the above ground speed is a more reasonable parameter to estimate the seismic scale.

Therefore, by using the seismometer 10 to generate an earthquake signal including the ground acceleration and the ground velocity, it is possible to minimize the malfunction of the earthquake notification and the distortion of the earthquake scale estimation, which may occur due to the high frequency noise, The reliability of seismic scale estimation can be improved by simultaneously considering both high frequency and low frequency seismic characteristics.

Here, the ground acceleration and the ground speed can be sampled by a predetermined number of times per unit time, and the sampled ground acceleration and the ground speed can be represented as a set of vectors having n _s sampling values within a unit time. For example, when the sampled ground acceleration is sampled n _s times for a unit time (1 second), a _is (t) = (a _i , a _{i +1} , ..., a _{i + ns -1} ) , And the sampled ground speed can be expressed as v _is (t) = (v _i , v _{i +1} , ..., v _{i + ns -1} ). At this time, the seismometer 10 can take an absolute value with respect to the sampled ground acceleration and the ground speed.

The seismometer 10 senses the seismic wave and can generate an earthquake signal including the sampled ground acceleration and the ground speed and transmit the seismic signal to the analysis unit 20.

The analysis unit 20 may calculate a characteristic parameter indicating the characteristic of the earthquake from the seismic signal and input the characteristic parameter into the artificial neural network function to derive the earthquake magnitude of the earthquake.

The characteristic parameters include a peak ground acceleration (PGA), a peak ground velocity (PGV), a cumulative absolute velocity (CAV), and a cumulative absolute displacement (CAD) .

The maximum ground acceleration corresponds to a maximum value among the ground acceleration (| a |) input during the earthquake duration (N) as shown in FIG. 2 (a) As shown, this means the maximum value among the ground speeds (| V |) input during the earthquake duration (N). The earthquake duration (N) is the duration of the seismic wave expected at the occurrence of an earthquake, and may be set between 3 and 40 seconds in general.

The cumulative absolute velocity may be obtained by integrating the input ground acceleration (| a |) by unit time and then integrating the integration result for the seismic duration (N). That is, the area between the consecutive two sampling values is calculated by using two consecutive sampling values and a sampling time interval from the ground acceleration values sampled by the predetermined number of times (n _s ) for a unit time, So that the integration result can be obtained. Here, when the maximum value of the ground acceleration (| a |) within the unit time is smaller than the reference ground acceleration a _th , the integration result may be excluded from the integration to remove noise included in the earthquake signal.

Specifically, the cumulative absolute velocity is

을 이용하여 구할 수 있으며, 여기서, CAV는 누적절대속도,

,

, t는 시간, a_th는 기준지반가속도, N은 지진지속시간, n_s는 단위시간당 샘플링수에 해당한다.

, Where CAV is the cumulative absolute velocity,

,

, t is the time, a _th is the reference ground acceleration, N is the seismic duration, and n _s is the number of samples per unit time.

The cumulative absolute displacement may be obtained by integrating the input ground speed (| v |) by unit time and then integrating the integration result during the earthquake duration (N). That is, the area between the consecutive two sampling values is calculated by using two consecutive sampling values and a sampling time interval among the ground speed values sampled by the predetermined number of times (n _s ) for the unit time, So that the integration result can be obtained. Here, if the maximum value of the ground speed (| v |) within the unit time is smaller than the reference ground speed v _th , the integration result may be excluded from the integration to remove noise included in the earthquake signal.

Specifically, the cumulative absolute displacement is

을 이용하여 구할 수 있으며, 여기서, CAD는 누적절대변위,

,

, t는 시간, v_th는 기준지반속도, N은 지진지속시간, n_s는 단위시간당 샘플링수에 해당한다.

, Where CAD is the cumulative absolute displacement,

,

, t is the time, v _th is the reference ground speed, N is the seismic duration, and n _s is the number of samples per unit time.

In addition, the characteristic parameter may further include Vs, which is an inherent value of the seismometer 10 according to the observation place where the seismic wave is observed. Vs represents an average shear wave velocity from the surface to a predetermined depth. In particular, Vs30, which is the average shear wave velocity from the ground to the underground 30 m, can be utilized. However, the present invention is not limited to this, and it is also possible to use an average shear wave velocity of 100 m underground or 10 m underground. Further, when there is no information on the average shear wave velocity from the surface to the underground 30 m, it is also possible to set a rough Vs30 value depending on whether the ground provided with the seismograph 10 is a soft ground or a rock ground. For example, Vs30 can be set to 300 m / sec for soft ground, and Vs30 can be set to 800 m / sec for rock ground.

Additionally, instead of the cumulative absolute velocity and the cumulative absolute displacement, the analysis unit 20 may calculate the BSPGA (Bracketed Summation of Peak Gound Acceleration Per Second) and the BSPGV (BSPGV) It can be calculated and used as a parameter. That is, the BSPGA and the BSPGV can be calculated by integrating the maximum value in each unit time instead of integrating the area, such as the cumulative absolute velocity and the cumulative absolute displacement.

In the case of the cumulative absolute velocity or the cumulative absolute displacement, the ground acceleration value and the ground speed value are required to be 100 sampling / sec or more per second. However, when using the BSPGA or BSPGV, only one information is used every second, . However, in terms of accuracy, it may be somewhat lower than when utilizing the cumulative absolute velocity and cumulative absolute displacement.

Specifically,

을 이용하여 상기 BSPGA를 구할 수 있으며, 상기 p_is는 앞서 설명한 바와 동일하므로, 각 단위시간 내의 최대 지반가속도값 중에서 기준지반가속도(a_th)보다 큰 지반가속도만이 적산될 수 있다. 따라서, 상기 기준지반가속도(a_th)보다 작은 지반가속도값은 상기 적산에서 제외할 수 있으며, 이로인해 상기 적산에서 잡음신호를 배제할 수 있다. 여기서, 상기 p_is대신에 p?_s를 입력하고, a_is(t) 대신에 v_is(t)를 입력하면 상기 BSPGV도 동일한 방식으로 구할 수 있다.

The BSPGA can be obtained by using BSPGA. Since p _is the same as described above, only the ground acceleration greater than the reference ground acceleration a _th among the maximum ground acceleration values within each unit time can be accumulated. Therefore, the ground acceleration less than the reference ground acceleration (a _th ) can be excluded from the integration, thereby eliminating the noise signal in the integration. Here, p instead of the p _{is? s} and input _{is is} (t) instead of a _is (t), the BSPGV can be obtained in the same manner.

The analysis unit 20 may calculate the seismic scale by inputting the calculated characteristic parameter into the artificial neural network function, and the artificial neural network function may be a two layer feed-forward (TLFF) function in particular.

Specifically, as shown in FIG. 3, the analysis unit 20 may form an artificial neural network. First, the artificial neural network includes an input layer including (P + 1) input nodes, a hidden layer including (Q + 1) intermediate nodes, and an output including one output node And an output layer. The artificial neural network may further include (P + 1) * (Q + 1) first weight set for each intermediate node associated with the input node, and (Q + +1) number of second weights may be included. Here, if the characteristic parameters are input correspondingly to each of the input nodes, an earthquake scale may be output to the output node according to the relationship between the input node, the intermediate node, the intermediate node, and the output node.

The artificial neural network function

와 같이 수식으로 나타낼 수 있다. 여기서, y는 출력노드, x_p는 p번째 입력노드, w⁽¹⁾ _qp는 p번째 입력노드와 q번째 중간노드 사이의 제1 가중치, w⁽²⁾ _q는 상기 q번째 중간노드와 출력노드 사이의 제2 가중치, x₀ = 1일 수 있다. 상기 g(x)는 상기 인공신경망 함수에 비선형 특성을 부여하기 위한 것으로서, 상기 수식에 나타난 바와 같이 logsig 함수를 활용할 수 있으며, 상기 logsig 함수 이외에 tansig함수를 이용하는 것도 가능하다.

As shown in Fig. Here, y is an output node, x _p the p-th input node, w ⁽¹⁾ _qp is p-th input node and the q-th first weight between the intermediate node, w ⁽²⁾ _q is the q-th intermediate node and the output node , X ₀ = 1. The g (x) is for imparting a nonlinear characteristic to the artificial neural network function. The logig function can be used as shown in the above equation. In addition to the logsig function, the tansig function can also be used.

In order to obtain an accurate earthquake scale using the artificial neural network function, it is important to set the first weight and the second weight. The weight can be set by training the artificial neural network function using seismic observation data including an earthquake signal and an earthquake scale of an earthquake observed in the past. That is, the first weight value and the second weight value can be set by adjusting the first weight value and the second weight value so as to derive the seismic scale indicated in the seismic observation data when the characteristic parameter represented by the seismic observation data is input .

For example, with the characteristic parameter,

를 상기 입력노드에 각각 입력하면, 상기 각각의 특성파라미터에 대하여 상기 제1 가중치 및 제2 가중치가 적용되어, y값인 지진규모가 도출될 수 있다. 여기서, lnPGA, lnCAV, lnPGV, lnCAD, lnVs는 각각 PGA, CAV, PGV, CAD에 대한 자연로그값이다.

Is input to the input node, the first weight and the second weight are applied to each of the characteristic parameters, and an earthquake scale of y value can be derived. Here, lnPGA, lnCAV, lnPGV, lnCAD, and lnVs are the natural logarithm values for PGA, CAV, PGV, and CAD, respectively.

In addition, as shown in FIG. 4, the analysis unit 20 can derive an earthquake scale using two artificial neural network functions (TLFF1 and TLFF2). That is, after setting a plurality of different reference ground accelerations a _th and a reference ground speed v _th , PGA, CAV, PGV, and CAD based on the reference ground acceleration and the reference ground velocity are calculated as an artificial neural network function TLFF1), it is possible to derive a plurality of earthquake magnitudes corresponding to each of the reference ground acceleration and the reference ground velocity.

Thereafter, the plurality of earthquake magnitudes are set as one candidate group, and the candidate group is input to the second artificial neural network function (TLFF2), so that an optimal earthquake scale can be derived from the earthquake magnitude in the candidate group.

The calculated earthquake magnitude can be changed according to the reference ground acceleration and the reference ground velocity and indirectly reflects the shape of the seismic envelope to the earthquake scale estimation according to the setting of the reference ground acceleration and the reference ground velocity, The degree of elimination of the included noise may be different. Therefore, it is possible to calculate the seismic scale more accurately by generating a plurality of candidate groups while changing the reference ground acceleration and the reference ground velocity, and then deriving an optimal seismic scale from the candidate groups.

In addition, since the ground speed having a strong characteristic to the ambient noise signal can be associated with the Sesimic Intensity quantified by the MMI (Modified Mercalli Intensity) class, the reference ground speed can be set in relation to the seismic intensity have. For example, it is possible to know the ground speed corresponding to the perceived acceleration of a person, and if the ground speed corresponding to the perceived acceleration is set to the reference ground speed, .

In addition, when the value of the accumulated absolute velocity is calculated as 0 according to the value of the reference ground acceleration, the analysis unit 20 calculates the average value of the earthquake magnitude calculated when the accumulated absolute velocity is not 0 Can be set to the earthquake scale. That is, when the accumulated absolute velocity is 0, it is impossible to calculate the value of lnCAV, so that the earthquake scale can not be calculated. Therefore, an average of the measured earthquake magnitudes when the remaining accumulated absolute velocity is not 0 can be obtained, and the earthquake magnitude when the accumulated absolute velocity is zero can be selected.

Also, if the value of the cumulative absolute displacement is calculated as 0 according to the value of the reference ground speed, the earthquake scale calculation may be terminated. If the cumulative absolute displacement is zero, it corresponds to the case where the ground speed exceeding the reference ground speed is not measured, so that it can be considered that there is no significant ground vibration in this case. Therefore, if the accumulated absolute displacement is zero, the earthquake scale calculation can be stopped.

5 is a flowchart showing an earthquake measuring method according to an embodiment of the present invention.

Referring to FIG. 5, an earthquake measuring method according to an embodiment of the present invention may include a seismic wave sensing step (S10), a characteristic parameter calculating step (S20), and an earthquake scale deriving step (S30).

Hereinafter, an earthquake measuring method according to an embodiment of the present invention will be described with reference to FIG.

The seismic wave sensing step S10 may generate an earthquake signal by sensing a seismic wave generated by the earthquake. In the sensing step S10, the seismic wave can be detected using a seismometer, and the ground acceleration and the ground speed can be measured from the seismic wave. Specifically, the ground acceleration and the ground speed can be sampled by a predetermined number of times per unit time. The ground speed may be calculated by integrating the ground acceleration.

The characteristic parameter calculation step (S20) can calculate a characteristic parameter indicating the characteristic of the earthquake from the seismic signal. The characteristic parameters include Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV), Cumulative Absolute Velocity (CAV), and Cumulative Absolute Displacement (CAD) can do.

2, the maximum ground acceleration (PGA) corresponds to a maximum value among the ground acceleration measured during the earthquake duration N, and the maximum ground speed PGV corresponds to the earthquake duration N, Of the measured ground velocity.

In addition, the cumulative absolute velocity and the cumulative absolute displacement may be calculated by the following equation

를 이용하여 각각 계산할 수 있다. 여기서, CAV는 누적절대속도, CAD는 누적절대변위,

,

,

, t는 시간, a_th는 기준지반가속도, v_th는 기준지반속도, N은 지진지속시간, n_s는 단위시간당 샘플링수에 해당한다.

Respectively. Where CAV is the cumulative absolute velocity, CAD is the cumulative absolute displacement,

,

,

, t is the time, a _th is the reference ground acceleration, v _th is the reference ground velocity, N is the seismic duration, and n _s is the number of samples per unit time.

In addition, the characteristic parameter may further include a characteristic value Vs, which is an inherent value according to the observation place where the seismic wave is observed. Instead of the cumulative absolute velocity and the cumulative absolute displacement, BSPGA (Bracketed Summation of Peak Gauss Acceleration per second) It is also possible to calculate BSPGV (BSPGV: Bracketed Summation of Peak Gound velocity per second) by using the characteristic parameter. Since Vs, BSPGA, and BSPGV have been described above, detailed description is omitted here.

In the earthquake scale derivation step S30, the characteristic parameter may be inputted to an artificial neural network (ANN) function to derive the earthquake magnitude of the earthquake. Here, the artificial neural network function may be a two layer feed-forward (TLFF) function.

Specifically, as shown in FIG. 3, an input layer including (P + 1) input nodes, a hidden layer including (Q + 1) intermediate nodes, and a single output (P + 1) * (Q + 1) first weights for each intermediate node associated with the input node, and a second weight for each intermediate node associated with the input node, (Q + 1) second weights for the output node associated with the intermediate node. Thereafter, each of the input nodes may correspond to the characteristic parameter, and the output node may correspond to the earthquake scale.

Wherein the artificial neural network function comprises:

와 같이 나타낼 수 있으며, y는 출력노드, x_p는 p번째 입력노드, w⁽¹⁾ _qp는 p번째 입력노드와 q번째 중간노드 사이의 제1 가중치, w⁽²⁾ _q는 상기 q번째 중간노드와 출력노드 사이의 제2 가중치, x₀ = 1일 수 있다. 따라서, 상기 입력노드에 상기 특성파라미터를 입력하면, 상기 인공신경망 함수에 의하여 지진규모가 계산되어 도출될 수 있다. 예를들어,

를 상기 입력노드에 각각 입력하여 상기 지진규모를 도출할 수 있다.

And can be expressed as, y is an output node, x _p the p-th input node, w ⁽¹⁾ _qp is p first weight between the second input node and the q-th intermediate node, w ⁽²⁾ _q is the q-th intermediate A second weight between the node and the output node, x ₀ = 1. Accordingly, when the characteristic parameter is input to the input node, the seismic scale can be calculated and derived by the artificial neural network function. E.g,

Can be input to the input node to derive the earthquake magnitude.

Here, in order to calculate an accurate earthquake scale through the artificial neural network function, it is important to set the first weight and the second weight. Therefore, it is possible to further include an artificial neural network learning step (not shown) for setting the first weight and the second weight. The artificial neural network learning step may be performed before the detection of the seismic wave.

Specifically, in the artificial neural network learning step, the artificial neural network function can be learned by using characteristic parameters and seismic scale included in the measured earthquake observation data, and through the learning, A first weight and a second weight for outputting a corresponding magnitude of the earthquake can be set. The seismic observation data may include an earthquake signal of an earthquake observed in the past and information on an earthquake scale of the earthquake.

In addition, the earthquake scale derivation step S30 may calculate a plurality of earthquake magnitudes corresponding to a plurality of different reference ground acceleration and reference ground velocity through a first process, and then, through the second process, You can choose the optimal earthquake scale from the scale.

Specifically, the first process may input characteristic parameters, which are respectively applied to a plurality of different reference ground acceleration and reference ground speed, to the artificial neural network function. As shown in the equation of the characteristic parameter, if the reference ground speed and the reference ground acceleration are different, the value of the characteristic parameter may be changed. Therefore, if the reference ground acceleration and the reference ground speed are set differently, a plurality of different earthquake magnitudes can be calculated, and a candidate group can be formed using the plurality of earthquake magnitudes.

Here, the value of the accumulated absolute velocity may be calculated as 0 according to the value of the reference ground acceleration. In this case, it is impossible to calculate the logarithm of the cumulative absolute velocity, so that the earthquake scale can not be calculated. However, since it is necessary to set an earthquake scale in order to derive an optimal earthquake scale value through the second process, an average value of the earthquake scale calculated when the cumulative absolute velocity is not 0 can be used as the earthquake scale.

Also, the value of the accumulated absolute displacement may be calculated as 0 according to the value of the reference ground speed. In this case, the earthquake scale calculation may be terminated. Particularly, when the cumulative absolute displacement value is 0, it corresponds to the case where the ground speed exceeding the reference ground speed is not measured, so that it can be considered that there is no significant ground vibration. Therefore, if the cumulative absolute displacement is zero, the earthquake scale calculation can be stopped.

In the second step, the candidate group may be input to the second artificial neural network function, and an optimal seismic scale value may be derived from the seismic scale included in the candidate group. Here, the second artificial neural network function may be set to derive an optimal seismic scale value from the seismic scale calculated according to the different reference ground acceleration and the reference ground velocity. That is, the second artificial neural network function may be learned to derive an earthquake scale closest to the actual earthquake magnitude stored in the earthquake observation data among a plurality of earthquake scale values calculated using the measured earthquake observation data .

FIGS. 6 to 8 are graphs showing the effect of the seismic scale estimation using the ground speed on the earthquake acceleration data (606 in Korea, 365 in Europe, and 2,883 in US NGA) A comparison chart and a graph to which an earthquake measuring method according to the example is applied. The number of samples per second of the seismic acceleration data is over 100, and a fourth order buttworth low frequency cutoff (cutoff frequency = 0.075Hz) filter is applied for the conversion into the velocity data and integrated in the time domain. The PGA, PGV, BSPGA, and BSPGV in FIG. 6 use the geometric mean of the calculated values for the two horizontal component time histories. The analysis was carried out for five cases (① ~ ⑤), and the application of the optimal combination method of earthquake scale to the reference ground acceleration (a _th , v _th ) The error reduction effect can be observed.

Specifically, comparing the cases of (1) and (2), (3) and (4), it is possible to confirm the error reduction effect according to the addition of the ground speed and to obtain an error reduction effect of 20% as shown in FIG. 6 and FIG. Also, comparing ⑤dhk ③, ⑤ and ④, it is possible to confirm the error reduction effect according to the application of a plurality of reference ground accelerations and the reference ground speed (a _th , v _th ). As shown in FIGS. 6 and 8, It is possible to obtain an effect of reducing the error further.

The present invention is not limited to the above-described embodiments and the accompanying drawings. It will be apparent to those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

10: Seismometer 20: Analysis department
S10: Seismic wave detection step S20: Characteristic parameter calculation step
S30: Steps to derive earthquake scale

Claims (16)

A seismic wave sensing step of sensing an earthquake wave generated by an earthquake to generate an earthquake signal;
A characteristic parameter calculation step of calculating a characteristic parameter indicating a characteristic of the earthquake from the seismic signal; And
And inputting the characteristic parameter into an artificial neural network (ANN) function to derive an earthquake magnitude of the earthquake.
The method according to claim 1, wherein the characteristic parameter calculation step
The characteristic parameters include a peak ground acceleration (PGA), a peak ground velocity (PGV), a cumulative absolute velocity (CAV), and a cumulative absolute displacement (CAD) Earthquake measurement method.
3. The method according to claim 2, wherein the characteristic parameter calculation step
A maximum earth acceleration and a maximum ground velocity are calculated by calculating a maximum value among the ground acceleration and the ground velocity measured during the earthquake duration time.
4. The method according to claim 3, wherein the characteristic parameter calculation step

Where CAV is the cumulative absolute velocity, CAD is the cumulative absolute displacement,

,

,

, where t is the time, a _th is the reference ground acceleration, v _th is the reference ground speed, N is the seismic duration, and n _s is the seismic measurement method for calculating the cumulative absolute velocity and cumulative absolute displacement using the number of samples per unit time .
2. The method of claim 1, wherein the step
Wherein the artificial neural network function is a two layer feed-forward (TLFF) function.
6. The method of claim 5, wherein the step
An output layer including an input layer including (P + 1) input nodes, a hidden layer including (Q + 1) intermediate nodes, and an output node Setting,
(P + 1) * (Q + 1) first weights for each intermediate node associated with the input node and a (Q + 1) second weight for output nodes associated with the intermediate node, ,
Wherein the characteristic parameter is associated with each of the input nodes, and the earthquake scale is associated with the output node, thereby forming an artificial neural network.
7. The method of claim 6, wherein the step

y is between the output node, x _p the p-th input node, w ⁽¹⁾ _qp is the first weight between the p-th input node and the q-th intermediate node, w ⁽²⁾ _q is the q-th intermediate node and the output node A second weight, x ₀ = 1, is used to derive the earthquake magnitude.
8. The method of claim 7,
Wherein the first and second weights are obtained by learning the artificial neural network function using characteristic parameters and seismic scale included in the measured seismic observation data and outputting the corresponding seismic scale when the characteristic parameter is input to the artificial neural network function, And an artificial neural network learning step of setting a second weight.
8. The method according to claim 7, wherein the earthquake scale derivation step

The input node is input to the input node to derive the seismic scale, and Vs is an average shear wave velocity from the surface to a predetermined depth.
8. The method according to claim 7, wherein the earthquake scale derivation step
A first step of inputting a characteristic parameter to which a plurality of different reference ground acceleration and a reference ground speed are respectively applied to the artificial neural network function and deriving a candidate group of the earthquake magnitude calculated corresponding to the plurality of reference ground acceleration and the reference ground speed ; And
And a second step of inputting the candidate group to a second artificial neural network function to derive an earthquake scale value from among the earthquake magnitudes included in the candidate group.
11. The method of claim 10, wherein the first step
And an average value of the seismic scale calculated when the cumulative absolute velocity is not 0 when the value of the cumulative absolute velocity is calculated as 0 according to the value of the reference ground acceleration.
11. The method of claim 10, wherein the first step
And terminating the calculation of the earthquake scale when the value of the cumulative absolute displacement is calculated as 0 according to the value of the reference ground speed.
11. The method of claim 10, wherein the second artificial neural network function
A method of measuring an earthquake of a plurality of seismic scale values according to a reference ground acceleration and a reference ground velocity generated by using the measured seismic observation data to derive the seismic scale closest to the actual earthquake magnitude stored in the seismic observation data.
A seismometer that generates an earthquake signal by sensing seismic waves generated by earthquakes; And
And an analyzing section for calculating a characteristic parameter representing a characteristic of the earthquake from the seismic signal and inputting the characteristic parameter to an artificial neural network function to derive an earthquake scale of the earthquake.
15. The apparatus of claim 14, wherein the analyzer
Wherein at least the maximum ground speed and the cumulative absolute displacement are included as the characteristic parameters.
15. The apparatus of claim 14, wherein the analyzer
A plurality of different reference ground accelerations and a reference ground speed are respectively inputted to the artificial neural network function to derive a candidate group of earthquake magnitudes calculated corresponding to the plurality of reference ground accelerations and the reference ground speed, And inputting the candidate group to a second artificial neural network function to derive one earthquake scale value among the earthquake magnitudes included in the candidate group.

Images