TW201805653A - Method, system, recording medium and computer program product for instant seismic analysis of story of building - Google Patents

Abstract

A method, a system, a recording medium and a computer program product are provided to perform response prediction of a story of a building. A regression model is ready established by earthquake information of previous earthquakes. When an earthquake occurs, earthquake information of the earthquake is received and input to the regression model, and the predicted peak floor acceleration may be determined.

Description

Earthquake analysis method, system, recording medium and computer program product on the building floor

The present invention relates to seismic analysis, and more particularly to an earthquake analysis of a building floor.

Taiwan is located in a region where earthquakes occur frequently. In the current earthquake, many buildings may have collapsed or damaged due to the inability to withstand the earthquake intensity. In the present day of September 21, many areas have damages caused by construction damage, which have a major impact on social stability and economic development. If the earthquake is not transmitted to the building, it will be early warning, and the residents of other buildings will carry out disaster prevention and avoidance actions. Earthquake disaster casualties were minimized. However, how to estimate the response of large buildings after the earthquake has always been the most concerned issue for engineers. However, earthquakes are sporadic events, which are short-lived (about tens of seconds to a few minutes) and are formally unreasonable and unpredictable. Therefore, accurate estimation of earthquake response is a very challenging target.

In the local earthquake strong earthquake alarm system, it is necessary to first understand the earthquake intensity at a certain location to further determine whether the area is early warning, and the Peak Ground Acceleration is a representative and important parameter for the earthquake intensity. Before distinguishing the magnitude of the earthquake intensity, it is necessary to estimate the maximum surface acceleration by using the seismic arrival wave characteristic parameters. Degree, its parameters are: three peak observations, Effective Predominant Period, and other integral values, including Integral of Squared Velocity and Cumulative Absolute Velocity.

However, all current strong earthquake early warning systems are used to predict the maximum acceleration of a station on the "ground". In fact, the maximum acceleration between the upper floors of the building and the maximum surface acceleration is not the same. The maximum acceleration of the upper floors should generally be larger and more dangerous than the maximum surface acceleration. For Taiwan with high buildings, it is not possible to fully predict the maximum surface acceleration.

Patent No. I467212 also provides a method for calculating the maximum floor acceleration. The earthquake real-time analysis system and its method and storage medium are constructed according to the requirements for the specific building dynamic model, and the earthquakes of the Central Meteorological Bureau and other institutions. The measurement data is used as an input factor to immediately calculate the structural deformation of the seismic wave. In addition, after accepting the seismic characteristic parameters of the first wave of the earthquake, it is necessary to select one of the multiple floor regression formulas according to the earthquake source location and the source distance, and calculate the amplification parameters of the specific floor, and calculate the specific Predicted seismic data for floors.

Compared with the above patents, the invention is relatively simple and fast, and the other is that the structural dynamic model of the specific building is not required to be specially constructed; the other is that the prediction data of the specific floor can be calculated without judging the source position and the source distance of the earthquake. However, despite the basis for omitting these two judgments, there are still fairly accurate prediction results in the embodiments of the present invention.

In view of the limitations of the prior art, the present invention provides an earthquake real-time analysis method for a building floor, and constructs a regression model through a computer system in advance to estimate a new land. The maximum floor acceleration of a particular floor of a particular building when the earthquake occurs, the method comprising: obtaining seismic information of a plurality of previous earthquakes in advance, the seismic information comprising: a plurality of seismic arrival wave characteristic parameters, and a plurality of buildings a structural period and a height of each floor; a maximum floor acceleration of each of the plurality of buildings when the plurality of previous earthquakes are obtained in advance; and a maximum floor acceleration of each of the plurality of buildings in advance through the computer system The regression information of the plurality of previous earthquakes constructs the regression model; when the new earthquake occurs, the seismic information of the new earthquake is detected within a specific time, and the seismic information of the new earthquake is substituted into the computer system. The regression model is constructed to estimate the maximum floor acceleration for that particular floor of the particular building.

The method for constructing the regression model by using the computer system to reconstruct the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes is a support vector regression machine, which is the plurality of previous earthquakes The seismic information and the maximum floor acceleration of each floor of the plurality of buildings conform to the following equation: Where f ( x ) is the maximum floor acceleration of the particular floor of the particular building when the new earthquake occurs, x ₁ to x _j are a plurality of vectors, x is a specific vector, α ₁ to α _j , β ₁ to β _j and b are constants derived from the support vector regression, the complex vector and the maximum floor acceleration of each floor of the plurality of buildings corresponding to the plurality of vectors, and the function k corresponds to a high One of the dimension feature spaces is a kernel function.

Wherein the kernel function is as follows: k ( x _i , x _j )=exp(-∥ x _i - x _j ||/2 σ ² ), where σ is a constant.

Where α ₁ to α _j , β ₁ to β _j are obtained according to solving a quadratic programming problem, which is as follows: limited by Where y ₁ to y _m are the maximum floor accelerations of the floors of the plurality of buildings corresponding to the plurality of vectors, and ε and C are constants.

The quadratic programming problem is obtained by introducing an objective function and introducing Lagrange multipliers, the objective function is as follows: limited by Where w is one of the vectors in the high dimensional feature space, ζ _i , And b is the variable of the objective function.

The seismic information corresponding to each of the plurality of vectors is a characteristic value of the plurality of seismic arrival wave characteristic parameters, a structural period of the plurality of buildings, and a height of each floor.

The specific vector is composed of the new seismic information.

The plurality of seismic arrival wave characteristic parameters include an absolute value of the absolute value of the acceleration (P _a ), an absolute value of the absolute value of the velocity (P _v ), an extreme value of the absolute value of the displacement (P _d ), an integral period of the effective master, and an integral of the square of the velocity. Quantity, as well as cumulative absolute speed.

The invention further provides a computer readable recording medium with a built-in program, which can be completed when the computer loads the program and executes it.

The invention further provides a computer program product for storing real-time analysis of earthquakes on a building floor, which can be completed when the computer is loaded into the computer program and executed.

The invention further provides an earthquake real-time analysis system for building floors, which constructs a regression model in advance through a computer system to estimate the maximum floor acceleration of a particular floor of a particular building when a new earthquake occurs, the system The utility model comprises: a storage unit for storing the regression model, wherein the regression model is constructed as follows: obtaining seismic information of a plurality of previous earthquakes, the seismic information comprising: a plurality of seismic arrival wave characteristic parameters, a structure of a plurality of buildings a period of the object and the height of each floor; the maximum floor acceleration of each floor of the plurality of buildings when the plurality of previous earthquakes are obtained; the maximum floor acceleration of each floor of the plurality of buildings and the earthquake of the plurality of previous earthquakes The information constructs the regression model; a transmission unit is configured to receive seismic information for detecting the new earthquake within a specific time when the new earthquake occurs; and an operation processing unit electrically connecting the storage unit with The transmission unit receives the seismic information of the new earthquake transmitted by the transmission unit, and the new Shock of the earthquake information is substituted into the regression model to estimate the maximum floor acceleration of the specific floor of this particular building.

The method for constructing the regression model by the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes is a support vector regression machine, which is the seismic information of the plurality of previous earthquakes and the complex number The maximum floor acceleration of each floor of a building conforms to the following equation: Where f ( x ) is the maximum floor acceleration of the particular floor of the particular building when the new earthquake occurs, x ₁ to x _j are a plurality of vectors, x is a specific vector, α ₁ to α _j , β ₁ to β _j and b are constants derived from the support vector regression, the complex vector and the maximum floor acceleration of each floor of the plurality of buildings corresponding to the plurality of vectors, and the function k corresponds to a high One of the dimension feature spaces is a kernel function.

Wherein the kernel function is as follows: k ( x _i , x _j )=exp(-∥ x _i - x _j ||/2 σ ² ), where σ is a constant.

Where α ₁ to α _j , β ₁ to β _j are obtained according to solving a quadratic programming problem, which is as follows: limited by Where y ₁ to y _m are the maximum floor accelerations of the floors of the plurality of buildings corresponding to the plurality of vectors, and ε and C are constants.

The quadratic programming problem is obtained by introducing an objective function and introducing Lagrange multipliers, the objective function is as follows: limited by Where w is one of the vectors in the high dimensional feature space, ξ _i , And b is the variable of the objective function.

The seismic information corresponding to each of the plurality of vectors is a characteristic value of the plurality of seismic arrival wave characteristic parameters, a structural period of the plurality of buildings, and a height of each floor.

The specific vector is composed of the new seismic information.

The plurality of seismic arrival wave characteristic parameters include an absolute value of the absolute value of the acceleration (P _a ), an absolute value of the absolute value of the velocity (P _v ), an extreme value of the absolute value of the displacement (P _d ), an integral period of the effective master, and an integral of the square of the velocity. Quantity, as well as cumulative absolute speed.

S10~S40‧‧‧ steps of an instant seismic analysis method for a building floor in an embodiment of the present invention

600‧‧‧ Earthquake Time Analysis System for Building Floors

610‧‧‧Operation Processing Unit

620‧‧‧ storage unit

630‧‧‧Transmission unit

1 is a schematic flow chart of an earthquake analysis method for a building floor in an embodiment of the present invention; FIG. 2 is a diagram showing the name and distribution position of a building and a measuring station in an embodiment of the present invention; In the embodiment of the invention, the sample data is used as a regression result of the training model; FIG. 3B is a regression result obtained by substituting the test sample data into the training model in an embodiment of the present invention; and FIG. 4A is an earthquake in the test sample data according to an embodiment of the present invention; Comparison of the maximum floor acceleration and earthquake of event one

FIG. 4B is a comparison of the maximum floor acceleration and the earthquake degree of the seismic event 2 in the test sample data according to an embodiment of the present invention.

Figure 4C is a comparison of the maximum floor acceleration and the earthquake of the seismic event 3 in the test sample data according to an embodiment of the present invention.

Figure 5 is a system block diagram of an earthquake real-time analysis system for a building floor in another embodiment of the present invention.

The invention relates to an earthquake real-time analysis system for a building floor, a method thereof and a storage medium. The methods disclosed in the following embodiments are described by the steps of the flowcharts, but the operations are not limited to the specific order of the steps shown in the flowcharts.

Please refer to FIG. 1 , which is a schematic diagram of an analysis process of estimating the maximum floor acceleration of a building floor according to an embodiment of the present invention. Please refer to FIG. 2 in combination, which is built in the embodiment of the present invention. The name and distribution location of the building and measurement station. The exact location of each station is shown in Table 1.

Step S10: The maximum floor acceleration of each floor of the plurality of buildings when the plurality of previous earthquakes are obtained in advance.

The invention selects 60 earthquake events with a seismic scale (M _L ) of 6.0 or more in the seismic data of the structural strong earthquake monitoring system of the Central Meteorological Administration, and a total of 788 seismic data. According to the meaning of the maximum floor acceleration, the maximum acceleration value of each channel's seismic duration is first taken, and then the channels of the same floor height of the strong earthquake are compared with each other to obtain the actual maximum floor acceleration of the floor. Among them, some stations have no vertical strong earthquake detectors on the ground, or other factors such as unclear elevation of the building's elevation map, so that the relevant seismic information cannot be fully known. Therefore, this study filtered one by one to remove the problematic seismic history data. The obtained earthquake events were converted into a total of 3031 sample data. The detailed information of the sample data of each building station is shown in Table 2.

Step S20: Acquire seismic information of a plurality of previous earthquakes in advance, the seismic information includes: a plurality of seismic arrival wave characteristic parameters, a structural period of the plurality of buildings, and a height of each floor.

The characteristic parameters of the multiple earthquakes of the previous earthquake can be obtained from the seismic history data in the vertical direction on the ground. The characteristics of the initial wave characteristics of the earthquake include: maximum acceleration (P _a ), maximum velocity (P _v ), maximum displacement (P _d ), effective period of excellence (Tc), integral of speed squared (IV2), and cumulative absolute speed (CAV).

From the time when the P wave arrives at the destination from the epicenter, the elapsed time is t _p seconds, and the observation time of the present invention is 3 seconds. The maximum acceleration, maximum speed and maximum displacement are the maximum values taken during the observation time. The effective period of excellence, the integral of the square of the speed, and the cumulative absolute speed are calculated by the following formula:

Where u(t) is the displacement in the vertical direction, It is the vertical velocity and ü(t) is the vertical acceleration.

According to the two related parameters of floor height and structure period, the present invention looks at the floor height of each strong station of each station one by one, and divides by ten as the structure period according to the maximum number of floors of the building, so each building station In an earthquake event, a set of six seismic arrival wave characteristic parameters and structure period, array floor height and actual maximum floor acceleration are obtained.

Step S30: The maximum floor acceleration and the plurality of seismic information are constructed by the computer system in advance to construct the regression model.

This study used 2274 sample data from the Central Meteorological Bureau structural strong earthquake monitoring system as representative seismic data to train the regression model of the support vector method. The support vector regression model is constructed by using eight seismic information such as the characteristics of the primary wave characteristics of the earthquake, the floor height and the structural period, and uses the "support vector regression method" to find the eight independent parameters and the maximum floor acceleration. The relationship between the floor height and the building cycle is fixed for a building, so the value of the value can be known before the earthquake occurs.

The main concept of the support vector regression method is: "Using a nonlinear projection function to project the original data space to a higher-dimensional feature space, in which it is easier to accurately predict the data using the hyperplane. Therefore, this study intends to predict the maximum floor acceleration of a building when an earthquake occurs by a model constructed by support vector regression.

If there is a test data ( x , y ) R , x represents the input data, each input data can be a plurality of parameters independent of each other, and y represents the true value corresponding to the entire test data. If a set of x values can directly or indirectly affect the y value, The regression vector method can be used to find the relational expression between the two. First let the regression function be:

In the above formula, w is the regression vector coefficient, which represents the direction of a hyperplane in H space, and b is the offset, representing the distance between this plane and the dot.

For each x _i , the error between f ( x _i ) and y _i is small and lies within the error tolerance ε , which means that f ( x ) can accurately predict y by x . However, in most applications, because of various factors such as noise and error, the input data points are often not easily placed in the correct area. Therefore, in order to support the regression vector machine with better interpretation ability, for the point that is not favorable for optimization, that is, points other than ε -insensitive, additional items ξ _i , To allow certain occasions to fall outside of ε .

In the above formula, ξ _i , For the error value of the data point falling within the tolerance error interval, it may be determined whether the difference between the predicted value and the target value is allowed to be greater than ε , and if the error is less than ε , the value is zero.

The v-support vector regression and the restricted form corrected by the convex optimization problem are as follows:

limited by

In the above formula, m is the number of test samples, C is the parameter that makes better choices in model complexity and allowable data error tolerance, and εv is the slack variables controlled by the coefficient v .

In order to understand the formula, the Lagrange multiplier method is introduced, and the Lagrangian multipliers α _i and β _{i are set} to obtain the dual problem, and a quadratic programming algorithm is used. ) Find the Lagrangian multiplier:

limited by

In the above formula, y ₁ to y _m are objective functions corresponding to the plurality of vectors, and ε and C are constants. k ( x _i , x _j ) is the kernel function, and the kernel function used here is the swath base function, as follows:

After the Lagrangian multiplier is obtained, the Karush-Kuhn-Tucker complementarity conditions (KKT) can be used to solve for w and b . Therefore, the decision function is:

In the above formula, j is the number of support vectors.

When training each parameter combination model, it is necessary to give the cost parameter C and the σ value required by the kernel function, and find the C value and σ value that can make the model achieve the best training result. This training model is the best model. Therefore, in the training of the model, the invention uses a large range of grid search to determine the optimal C and σ values. The common range of σ values is 2 ^-1 ~ 2 ^-12 , and the C value is commonly used. The range is 2 ⁵ ~ 2 ¹⁵ . After comparison within the defined range, the C and σ values corresponding to the square root error of the final model are the best parameters. Since the smaller the σ value, the more computation time is required. Therefore, this study uses the method of segmentation to find the best parameters. First, find the σ value from 2 ^-12 ~ 2 ^-6 , such as the best solution. For a given range of bounds, such as σ _best = 2 ^-6 , then change the σ value to a range of 2 ^-6 ~ 2 ^-1 to more efficiently find the best parameters.

Step S40: When the new earthquake is about to arrive, the plurality of seismic information of the new earthquake is substituted into the regression model constructed by the computer system to estimate the maximum floor acceleration of the specific floor.

Substituting the seismic characteristic parameters and the structure-related parameters of the first wave of an earthquake into the support vector regression model, the maximum floor acceleration of a particular floor can be calculated.

In the example of the present invention, 2274 sample materials are used as the training model of the support vector regression method, and 757 sample data are substituted into the training model to obtain the predicted value of the maximum floor acceleration. In order to examine the regression of the model, the present invention uses the seismic intensity classification of the Central Meteorological Administration as the standard, and the detailed classification criteria are shown in Table 3. If the earthquake intensity corresponding to the predicted value of the maximum floor acceleration and the actual value of the maximum floor acceleration are between the positive and negative levels, it is judged that the data training result is better.

Please refer to FIG. 3A, which is a regression result of 2274 sample data as a training model according to an embodiment of the present invention, wherein the optimal cost parameter C obtained by the lattice search method is 4096, and the optimal σ value is 0.0156. The squared correlation coefficient of the regression is 0.89813; the 3B is the regression result obtained by substituting 757 test sample data into the training model, wherein the square correlation coefficient of the regression is 0.319626, the predicted value of the maximum floor acceleration and the maximum floor acceleration The comparison of the actual values, the accuracy of the seismic grading between the positive and negative levels is 95.51%.

In order to examine the practicability of the present invention, the present invention randomly selects three seismic events. The detailed seismic event information is shown in Table 4. The historical seismic data of each seismic event is substituted into the support vector regression model, and the prediction results and actual results are obtained. Discuss it. 4A, 4B, 4C are After the seismic characteristic parameters and the building characteristic parameters of the three seismic events are substituted into the support vector regression model, the predicted seismic data obtained is compared with the actual earthquake-stimulated situation, including the maximum floor acceleration and the corresponding seismicity. It can be seen from the three comparison graphs that except for the first earthquake event, the estimated value of the height of 20.95 meters is smaller than the actual value, the other estimated maximum floor accelerations are greater than the actual value, and the earthquake prediction value of each event. Both are equal to the actual earthquake or a large one. Therefore, it can be inferred that the method of the present invention is conservative and precise.

Please refer to FIG. 5, which is a block diagram of a system architecture of an earthquake real-time analysis system for a building floor according to an embodiment of the present invention. The seismic real-time analysis system 600 of the building floor can basically be any type of computer system as long as the seismic real-time analysis method of various buildings described in the foregoing embodiments can be smoothly performed. The earthquake real-time analysis system 600 of the building floor mainly includes an operation processing unit 610, a storage unit 620, and a transmission unit 630.

The storage unit 620 is configured to store digital data, and may include a system memory used by the operation processing unit 610 in a broad sense, a built-in volatile or non-volatile memory device, or even a network that can be accessed by connecting with the system 600. The storage unit 620 is configured to store the regression model constructed in the foregoing step S30.

The transmission unit 630 can be any wired or wireless network transmission device, as long as the system 600 can be connected to the analysis system of the measurement station to transmit data, and the receiving measurement station detects the new ground. Earthquake information.

The operation processing unit 610 is electrically connected to the storage unit 620 and the transmission unit 630, and can receive the seismic information of the new earthquake transmitted by the transmission unit 630, and substitute the seismic information of the new earthquake into the regression model stored in the storage unit 620. The maximum floor acceleration of a particular floor of a particular building is estimated by an operation; the arithmetic processing unit 610 can be implemented by a central processing unit (CPU), a microprocessor, an integrated circuit, or a wafer.

In order to achieve the effect of the earthquake warning, after estimating the maximum floor acceleration of a particular floor of a particular building, the predicted seismic data or the indicated hazard level and evacuation indication of the particular floor may be output visually or acoustically.

In summary, the present invention has the following features:

(1) Instant and fast: Through the support vector regression model of regression analysis, the fast-calculated calculation of the predicted seismic data of a specific floor is relatively faster than the mechanical analysis process of structural dynamics, which is faster in calculation and conforms to the rapid response assessment of buildings. Strong earthquake warning needs.

(2) High accuracy: In one example of the present invention, 2274 sample data is used as the training model, 757 test sample data are substituted into the training model, and the seismic intensity classification of the Central Meteorological Administration is used as the standard, and the final seismic grading is between the positive and negative levels. The rate of accuracy is as high as 95.51%.

However, the above is only a preferred embodiment of the present invention, and the scope of the invention is not limited thereto; therefore, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent.

S10~S40‧‧‧ steps of an instant seismic analysis method for a building floor in an embodiment of the present invention

Claims (18)

An earthquake real-time analysis method for building floors, in which a regression model is constructed in advance through a computer system to estimate the maximum floor acceleration of a particular floor of a particular building when a new earthquake occurs, the method comprising: pre-acquisition The maximum floor acceleration of each of the plurality of buildings in the plurality of previous earthquakes; obtaining seismic information of a plurality of previous earthquakes in advance, the seismic information comprising: a plurality of seismic arrival wave characteristic parameters, and a structure of the plurality of buildings a period of the object and a height of each floor; the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes are preliminarily constructed by the computer system to construct the regression model; when the new earthquake occurs, Detecting the seismic information of the new earthquake within a specific time, and substituting the seismic information of the new earthquake into the regression model constructed by the computer system to estimate the maximum floor acceleration of the specific floor of the specific building . The method for real-time analysis of earthquakes of a building floor according to claim 1, wherein the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes are constructed in advance by the computer system to construct the regression model The method is a support vector regression machine, which is to match the seismic information of the plurality of previous earthquakes and the maximum floor acceleration of each floor of the plurality of buildings to the following equation: Where f ( x ) is the maximum floor acceleration of the particular floor of the particular building when the new earthquake occurs, x ₁ to x _j are a plurality of vectors, x is a specific vector, α ₁ to α _j , β ₁ to β _j and b are constants derived from the support vector regression, the complex vector and the maximum floor acceleration of each floor of the plurality of buildings corresponding to the plurality of vectors, and the function k corresponds to a high One of the dimension feature spaces is a kernel function. An earthquake real-time analysis method for a building floor as claimed in claim 2, wherein the kernel function is as follows: k ( x _i , x _j )=exp(−∥ x _i - x _j ∥/2 σ ² ), where σ is A constant. An earthquake real-time analysis method for a building floor as claimed in claim 2, wherein α ₁ to α _j , β ₁ to β _j are obtained according to a quadratic programming problem, and the secondary planning problem is as follows: limited by Where y ₁ to y _m are the maximum floor accelerations of the floors of the plurality of buildings corresponding to the plurality of vectors, and ε and C are constants. The method for real-time analysis of earthquakes on a building floor as described in claim 4, wherein the quadratic programming problem is obtained by introducing an objective function and introducing a Lagrange multipliers, the objective function is as follows: limited by Where w is one of the vectors in the high dimensional feature space, ξ _i , And b is the variable of the objective function. The seismic real-time analysis method for a building floor according to claim 2, wherein the seismic information corresponding to each of the plurality of vectors is a characteristic parameter of the plurality of earthquakes, and a structure of the plurality of buildings The characteristic value of the object period and the height of each floor. The seismic real-time analysis method for a building floor according to claim 2, wherein the specific vector is composed according to the new seismic information. The method for real-time analysis of earthquakes on a building floor as claimed in claim 1, wherein the plurality of seismic arrival wave characteristic parameters include an absolute value of the acceleration absolute value (P _a ), an absolute value of the absolute value of the speed (P _v ), and an absolute value of the displacement absolute value. Value (P _d ), effective master cycle, integral amount of velocity squared, and cumulative absolute velocity. An earthquake real-time analysis system for building floors, in which a regression model is constructed in advance through a computer system to estimate the maximum floor acceleration of a particular floor of a particular building when a new earthquake occurs, the system comprising: a storage a unit for storing the regression model, the regression model being constructed as follows: obtaining a maximum floor acceleration of each floor of the plurality of buildings when the plurality of previous earthquakes are obtained; acquiring seismic information of the plurality of previous earthquakes, the seismic information The method includes: a plurality of seismic arrival wave characteristic parameters, a structural period of the plurality of buildings, and a height of each floor; constructing the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes a regression unit, configured to receive seismic information for detecting the new earthquake in a specific time when the new earthquake occurs; and an operation processing unit electrically connecting the storage unit and the transmission unit, Receiving transfer order The seismic information of the new earthquake transmitted by Yuan, and the seismic information of the new earthquake is substituted into the regression model to estimate the maximum floor acceleration of the particular floor of the particular building. The earthquake real-time analysis system for a building floor according to claim 9, wherein the maximum floor acceleration of each floor of the plurality of buildings and the seismic information of the plurality of previous earthquakes are constructed by the regression model is a support vector regression The machine is configured to conform the earthquake information of the plurality of previous earthquakes and the maximum floor acceleration of each floor of the plurality of buildings to the following equation: Where f ( x ) is the maximum floor acceleration of the particular floor of the particular building when the new earthquake occurs, x ₁ to x _j are a plurality of vectors, x is a specific vector, α ₁ to α _j , β ₁ to β _j and b are constants derived from the support vector regression, the complex vector and the maximum floor acceleration of each floor of the plurality of buildings corresponding to the plurality of vectors, and the function k corresponds to a high One of the dimension feature spaces is a kernel function. An earthquake real-time analysis system for a building floor as claimed in claim 10, wherein the kernel function is as follows: k ( x _i , x _j )=exp(−∥ x _i - x _j ∥/2 σ ² ), where σ is A constant. An earthquake real-time analysis system for a building floor as claimed in claim 10, wherein α ₁ to α _j , β ₁ to β _j are obtained according to solving a quadratic programming problem, the secondary planning problem being as follows: limited by Where y ₁ to y _m are the maximum floor accelerations of the floors of the plurality of buildings corresponding to the plurality of vectors, and ε and C are constants. An earthquake real-time analysis system for a building floor as claimed in claim 12, wherein the quadratic programming problem is obtained by introducing an objective function and introducing a Lagrange multipliers, the objective function is as follows: limited by Where w is one of the vectors in the high dimensional feature space, ζ _i , And b is the variable of the objective function. The seismic real-time analysis system for a building floor according to claim 10, wherein the seismic information corresponding to each of the plurality of vectors is a characteristic parameter of the plurality of earthquakes, and a structure of the plurality of buildings The characteristic value of the object period and the height of each floor. An earthquake real-time analysis system for a building floor as claimed in claim 10, wherein the specific vector is composed of the new seismic information. The earthquake real-time analysis system for a building floor according to claim 9, wherein the plurality of seismic arrival wave characteristic parameters include an absolute value of the acceleration absolute value (P _a ), an absolute value of the absolute value of the speed (P _v ), and an absolute value of the displacement. Extremum (P _d ), effective master period, integral amount of velocity squared, and cumulative absolute velocity. A computer readable recording medium for storing a program, and when the computer loads the program and executes it, the method as claimed in claims 1 to 8 can be completed. A computer program product for storing an earthquake for real-time analysis of a building floor, after the computer is loaded into the computer program and executed, the method described in claims 1 to 8 can be completed.