Earthquake frequency prediction method based on delay correlation

By acquiring earthquake catalog data of the fault system, constructing earthquake frequency time series and calculating delay cross-correlation coefficients, and building a regression model, the accuracy and quantifiability issues of earthquake frequency prediction in existing technologies are solved, and high-precision future earthquake frequency prediction is achieved.

CN121878773APending Publication Date: 2026-04-17CHENGDU UNIVERSITY OF TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511932365.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-19
Publication Date
2026-04-17

AI Technical Summary

Technical Problem

Existing technologies lack quantitative statistical prediction models based on the correlation between fault-induced seismic delay and frequency, making it impossible to make high-precision and quantifiable predictions of the frequency of seismic activity within fault systems.

Method used

By acquiring seismic catalog data of the fault system region, assigning seismic events to corresponding fault zones, statistically analyzing seismic frequencies, constructing a seismic frequency time series, calculating the delay cross-correlation coefficient, building a regression model, selecting independent variables, and outputting predicted values ​​for future seismic frequencies.

Benefits of technology

It achieves high-precision earthquake frequency prediction based on the correlation between fracture-induced earthquake delay, providing a reliable data foundation and model support, and overcoming the shortcomings of existing methods in combining quantitative prediction with mechanism.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121878773A_ABST
    Figure CN121878773A_ABST
Patent Text Reader

Abstract

The invention provides an earthquake frequency prediction method based on delay correlation, and belongs to the field of earthquake frequency prediction, and the method comprises the steps: obtaining earthquake catalog data in a fracture system region, generating a complete earthquake catalog, carrying out the statistics of an earthquake frequency time sequence based on a preset earthquake magnitude range and a time unit, and calculating the delay cross correlation coefficient of the earthquake frequency time sequence, selecting a maximum delay cross correlation coefficient and a corresponding lag order from the seismic frequency time sequence according to the delay cross correlation coefficients; and according to the seismic frequency time sequence, the maximum delay cross correlation coefficient and the corresponding lagging order, constructing a seismic frequency time sequence delay correlation influence path network, through stepwise regression, obtaining a regression model, and screening out a regression model with the lagging order less than 0 to carry out seismic frequency prediction. According to the method, the problem that high-precision and quantifiable prediction cannot be carried out on the seismic activity frequency in a fracture system due to the lack of quantitative statistical prediction based on fracture seismic delay correlation in the prior art is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of earthquake frequency prediction, and particularly relates to an earthquake frequency prediction method based on delay correlation. Background Technology

[0002] Tectonic earthquakes are a type of natural earthquake, caused by faulting, specifically by the formation of faults in the lithosphere. They are also called fault earthquakes or rupture earthquakes. The mechanism of tectonic earthquakes involves the collision, compression, or stretching of tectonic plates, leading to increased stress and deformation. When this stress or deformation exceeds the tectonic plate's bearing capacity, the tectonic plate suddenly shifts or fractures, releasing the accumulated energy rapidly within a very short time, causing ground vibrations and forming an earthquake. During this process, stress migrates or transforms through interactions between adjacent tectonic plates or faults, triggering or inhibiting earthquakes on other nearby faults, with a certain time delay. In other words, seismic activity along adjacent fault zones should exhibit correlation and lag.

[0003] Current research on the interactions of faults at various scales and their correlation with seismic activity mainly focuses on tectonic dynamics, monitoring plate movement rates and directions, Coulomb stress, and scattering intensity. Based on the Coulomb rupture criterion and the spatial distribution of faults, it analyzes the patterns of mutual influence between adjacent faults and their seismic activity. While some studies have employed numerical simulation methods such as finite element methods, cellular automata, and nonlinear dynamic networks, there are relatively few studies that use seismic events to infer the correlation of large-scale fault zone seismic activity. Only geostatistical methods are used to estimate the clustering of seismic spatiotemporal distributions, or similarity measures and matching methods are used to calculate and classify the similarity of regional seismic sequences based on the convergence of time series, suggesting a correlation among similar regional seismic activities. However, these methods do not truly reveal the correlation of regional seismic activity; similarity is not entirely equivalent to correlation. Furthermore, the spatiotemporal clustering characteristics of earthquakes only reflect the possibility of correlation among the seismic activity of adjacent faults, not a certainty. Therefore, quantitative statistical prediction of earthquakes based on the correlation of fault seismic delays is currently lacking. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides an earthquake frequency prediction method based on delay correlation, which solves the problem that existing technologies lack quantitative statistical prediction models based on the delay correlation of fault seismic activity, resulting in the inability to make high-precision and quantifiable predictions of the frequency of seismic activity within fault systems.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a method for predicting earthquake frequency based on delay correlation, comprising the following steps: Obtain earthquake catalog data within the fault system region, assign all earthquake events to the corresponding fault zones in the fault system based on the earthquake catalog data, and count the frequency of earthquakes of magnitude greater than or equal to each fault zone to obtain a complete earthquake catalog. Based on the preset magnitude range and time unit, the earthquake frequency of each fault zone in the complete earthquake catalog is statistically analyzed and arranged in chronological order to obtain the earthquake frequency time series. Calculate the delay cross-correlation coefficient between each pair of data in the earthquake frequency time series, and select the maximum delay cross-correlation coefficient and its corresponding lag order from the earthquake frequency time series based on the delay cross-correlation coefficient; Based on the earthquake frequency time series, the maximum delay cross-correlation coefficient and its corresponding lag order, a delay-related influence path network of the earthquake frequency time series was constructed. Based on different combinations of independent and dependent variables, the stepwise method was used to screen the independent variables and obtain the regression model. Select regression models with lag order less than 0 from the regression models, input the known independent variable data, and output the predicted value of future earthquake frequency.

[0006] To address the lack of effective earthquake frequency prediction based on the correlation between fault-induced seismicity delay in existing technologies, this invention provides an earthquake frequency prediction method based on the correlation between fault-induced seismicity delay. This method determines the minimum complete magnitude and generates a complete earthquake catalog, comprehensively recording historical earthquake data within the fault system and data from different fault zones, providing a data foundation for subsequent calculations and comparisons. By constructing an earthquake frequency time series, earthquake events are transformed into time series data, quantifying seismic activity. The method characterizes the intensity and time difference of the mutual influence of earthquake events in different fault zones, as well as the correlation between earthquake transmission and triggering delays between fault zones, by calculating the delay cross-correlation coefficient and its corresponding lag order. Finally, by constructing and filtering a regression model, inputting known independent variable data (i.e., known earthquake frequency time series), and outputting predicted values ​​for future earthquake frequencies, this method achieves earthquake frequency prediction based on the correlation between fault-induced seismicity delay.

[0007] Furthermore: the obtained complete earthquake catalog specifically includes: Obtain earthquake catalog data within the fault system region, and assign all earthquake events to the corresponding fault zones in the fault system based on the earthquake locations in the earthquake catalog data; The frequency of earthquakes with magnitudes greater than or equal to each fault zone was statistically analyzed, and the Gutenberg-Richard relation was fitted using the least squares method to obtain the theoretical relationship between earthquake frequency and magnitude, i.e., the logarithmic fitted value of earthquake frequency. The logarithmic fitted values ​​of earthquake frequencies for each fault zone are compared with the measured logarithmic values ​​of earthquake frequencies. When the measured logarithmic values ​​of earthquake frequencies for each fault zone are all less than the logarithmic fitted values, the magnitude corresponding to the logarithmic fitted values ​​of earthquake frequencies is taken as the minimum complete magnitude. In the earthquake catalog data, all earthquakes with magnitudes smaller than the minimum complete magnitude are filtered out to obtain a complete earthquake catalog.

[0008] Furthermore, the expression for the logarithmic fitted value of the earthquake frequency is as follows:

[0009] in, The logarithmic fitted value of earthquake frequency represents the theoretical relationship between earthquake frequency and magnitude. For a certain magnitude, For a certain magnitude Earthquake frequency of magnitudes 1 and below and These are all coefficients that need to be estimated. This represents the overall seismic activity within the statistical spatiotemporal range. This represents the activity of high-magnitude earthquakes within the spatiotemporal range of the statistics.

[0010] The further beneficial effects mentioned above are as follows: This invention uses the least squares method to fit the Gutenberg-Richter relationship and compares the theoretical and measured logarithms of the frequencies of each magnitude to determine the completeness of earthquake events in each fault zone. Incomplete earthquake events are removed to obtain a complete earthquake catalog. This invention identifies and filters out small-magnitude events that are not fully recorded based on statistical laws, and constructs a statistically complete earthquake catalog, providing a reliable data foundation for subsequent statistical delay correlation analysis and earthquake frequency prediction.

[0011] Furthermore: the obtained earthquake frequency time series specifically includes: Based on the goals and analysis requirements of earthquake prediction, the magnitude range and time unit of all fault zones are uniformly set to obtain the preset magnitude range and time unit. Based on the preset magnitude range and time unit, the number of earthquake events in each fault zone within the time unit is counted in the complete earthquake catalog to obtain the original earthquake frequency sequence of each fault zone. Based on the original earthquake frequency sequence, the earthquake frequency time series is obtained by arranging them in chronological order and normalizing them.

[0012] Furthermore: the expression for the delayed cross-correlation coefficient is as follows:

[0013] in, To delay the cross-correlation coefficient, To find the covariance, for time Sequence value, for time Sequence value, for standard deviation for standard deviation and All data are from earthquake frequency time series. To delay the lag order corresponding to the cross-correlation coefficient, For time; select from the earthquake frequency time series that exceeds the specified time. The maximum delay cross-correlation coefficient and its corresponding lag order for the interval of 2 times the standard deviation.

[0014] The further beneficial effects mentioned above are as follows: This invention sets a unified magnitude range and time unit for all fault zones. The magnitude range is the range of earthquakes whose frequency of occurrence needs to be predicted, such as a magnitude range of 3.0 to 4.4, predicting the frequency of earthquakes within this magnitude range. The time unit is such as year, quarter, and month. By setting the magnitude range and time unit, the original earthquake frequency sequence is statistically obtained from a complete earthquake catalog. Normalization eliminates the magnitude difference in seismic activity levels between different fault zones, making it easier to compare the coordinated change trends of earthquakes. This makes earthquakes in different fault zones comparable, providing a foundation for subsequent earthquake frequency prediction based on the inter-earthquake delay correlation.

[0015] Furthermore: the obtained regression model specifically includes: Using earthquake frequency time series as nodes, maximum delay cross-correlation coefficients as edges, and their corresponding lag orders as the direction and weight of the edges, a delay-related influence path network of earthquake frequency time series is constructed; the edges pointing from one earthquake frequency time series to another are marked as pointing from the independent variable series to the dependent variable series. Traverse the edges of the earthquake frequency time series delay correlation path network to obtain several paths, and generate a combination for each path with the lagged form of the independent variable sequence as the independent variable and the current value of the dependent variable sequence as the dependent variable, to obtain different combinations of independent and dependent variables. Based on different combinations of independent and dependent variables, the stepwise method is used to screen the independent variables and obtain the regression model.

[0016] The further beneficial effects mentioned above are as follows: This invention transforms the abstract statistical correlation of data into a network model by constructing a time series delay correlation influence path network of earthquake frequency, giving it physical interpretability, and obtaining a model with excellent earthquake frequency prediction performance and physical basis through stepwise regression.

[0017] The beneficial effects of this invention are as follows: By constructing a delayed correlation path network of earthquake frequency time series using a one-way connected graph, and sorting out different combinations of independent and dependent variables that can be used to construct a regression model, the independent variables are screened using stepwise regression, thereby fitting a regression model in which the overall equation and regression coefficients pass the significance test and have weak multicollinearity. This invention achieves a quantitative and structured model expression of the delayed correlation between earthquake frequency time series, providing reliable and interpretable model support for earthquake frequency prediction based on statistical laws, and overcoming the shortcomings of existing methods in combining quantitative prediction with mechanism. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a method for predicting earthquake frequency based on delay correlation. Figure 2 The GR relationship for the target fracture system; Figure 3 This is a network diagram illustrating the path of influence of the correlation between earthquake frequency and time series. Detailed Implementation

[0019] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0020] Example 1 like Figure 1 The diagram shown is a flowchart of an earthquake frequency prediction method based on delay correlation. This invention provides an earthquake frequency prediction method based on delay correlation, comprising the following steps: Obtain earthquake catalog data within the fault system region, assign all earthquake events to the corresponding fault zones in the fault system based on the earthquake catalog data, and count the frequency of earthquakes of magnitude greater than or equal to each fault zone to obtain a complete earthquake catalog. Based on the preset magnitude range and time unit, the earthquake frequency of each fault zone in the complete earthquake catalog is statistically analyzed and arranged in chronological order to obtain the earthquake frequency time series. Calculate the delay cross-correlation coefficient between each pair of data in the earthquake frequency time series, and select the maximum delay cross-correlation coefficient and its corresponding lag order from the earthquake frequency time series based on the delay cross-correlation coefficient; Based on the earthquake frequency time series, the maximum delay cross-correlation coefficient and its corresponding lag order, a delay-related influence path network of the earthquake frequency time series was constructed. Based on different combinations of independent and dependent variables, the stepwise method was used to screen the independent variables and obtain the regression model. Select regression models with lag order less than 0 from the regression models, input the known independent variable data, and output the predicted value of future earthquake frequency.

[0021] This invention, from the perspective of analyzing earthquake event record data, utilizes readily available earthquake catalog data within fault system regions. After standardizing magnitude, seismogenic location matching, and completeness analysis, reasonable time units and magnitude ranges are selected. Based on the time units and magnitude ranges, earthquake frequencies of different fault zones are statistically analyzed to construct earthquake frequency time series. The lag correlation coefficients of each earthquake frequency series are calculated. Based on the largest statistically significant lag correlation coefficient between earthquake frequency time series, a network of sequence lag correlation influence paths is analyzed and constructed, obtaining various combinations of independent and dependent variables. A stepwise regression model is constructed, and finally, a regression model with lag order less than 0 for independent variables is selected to predict and calculate earthquake frequencies. The goodness of fit R of the regression model constructed in this invention is [not specified]. 2 It can achieve a maximum accuracy of 100%, a perfect fit, and realize accurate prediction of earthquake frequency, meeting the need to improve the accuracy of earthquake frequency prediction. At the same time, this invention uses earthquake catalog data to construct a high-precision quantitative earthquake frequency prediction model based on the delay correlation of fault-induced earthquakes, filling the gap in the lack of earthquake frequency prediction based on the delay correlation of fault-induced earthquakes.

[0022] In one embodiment of the present invention, due to varying monitoring capabilities of different seismic stations, records of minor earthquake events are incomplete in the original earthquake catalog data, and the data integrity is inconsistent between different fault zones. Therefore, incomplete earthquakes need to be removed to improve data quality. The process is as follows: Earthquake catalog data within the fault system region is obtained; all earthquake events are assigned to the corresponding fault zones in the fault system based on the earthquake catalog data; and the frequency of earthquakes of magnitude greater than or equal to each magnitude is counted for each fault zone to obtain a complete earthquake catalog. Specifically, this includes: Obtain earthquake catalog data within the fault system region. Based on the earthquake locations in the earthquake catalog data, i.e., longitude and latitude, assign all earthquake events to the corresponding fault zones in the fault system. The earthquake catalog data includes information such as the time of occurrence, longitude, latitude, and magnitude. To facilitate the subsequent division of earthquake magnitude ranges, the same method of determination must be used for the magnitude. The frequency of earthquakes with magnitudes greater than or equal to each fault zone was statistically analyzed, and the Gutenberg-Richard relation was fitted using the least squares method to obtain the theoretical relationship between earthquake frequency and magnitude, i.e., the logarithmic fitted value of earthquake frequency; the expression for the logarithmic fitted value of earthquake frequency is as follows:

[0023] in, The logarithmic fitted value of earthquake frequency represents the theoretical relationship between earthquake frequency and magnitude. For a certain magnitude, For a certain magnitude Earthquake frequency of magnitudes 1 and below and These are all coefficients that need to be estimated. This represents the overall seismic activity within the statistical spatiotemporal range. This represents the activity of high-magnitude earthquakes within the spatiotemporal range of the statistics.

[0024] The logarithmic fitted values ​​of earthquake frequencies for each fault zone are compared with the measured logarithmic values ​​of earthquake frequencies. When the measured logarithmic values ​​of earthquake frequencies for each fault zone are all less than the logarithmic fitted values, the magnitude corresponding to the logarithmic fitted values ​​is determined. As the minimum complete magnitude; In the earthquake catalog data, all earthquakes with magnitudes smaller than the minimum complete magnitude are filtered out to obtain a complete earthquake catalog, ensuring that the earthquake frequencies of the fault zones are statistically comparable.

[0025] In one embodiment of the present invention, due to the differences in the magnitude of seismic activity levels among different fault zones and the uneven temporal distribution of historical data, further processing of the complete earthquake catalog is required to improve the comparability of the data. Based on a preset magnitude range and time unit, the earthquake frequencies of each fault zone in the complete earthquake catalog are statistically analyzed and arranged chronologically to obtain an earthquake frequency time series, thus transforming discrete earthquake events into a continuous sequence. Specifically, this includes: Based on the goals and analysis requirements of earthquake prediction, the magnitude range and time unit of all fault zones are uniformly set to obtain the preset magnitude range and time unit. Among them, the magnitude range can be set to less than 3.0, between 3.0 and 4.4, and greater than 4.4. The time unit is set according to the amount of earthquake event data in the complete earthquake catalog of each fault zone, including year, quarter, month, etc. Based on the preset magnitude range and time unit, the number of earthquake events in each fault zone within the time unit is counted in the complete earthquake catalog to obtain the original earthquake frequency sequence of each fault zone. Based on the original earthquake frequency sequence, the sequence is arranged in chronological order and normalized to obtain the earthquake frequency time series. The normalization process eliminates the magnitude differences in seismic activity levels between different fault zones, making it easier to compare the coordinated change trends of the earthquake frequency time series.

[0026] In one embodiment of the present invention, in order to extract meaningful delay correlations from earthquake frequency time series, the present invention calculates the delay cross-correlation coefficient between each pair of data in the earthquake frequency time series, and selects the maximum delay cross-correlation coefficient and its corresponding lag order from the earthquake frequency time series based on the delay cross-correlation coefficient; the expression for the delay cross-correlation coefficient is as follows:

[0027] in, To delay the cross-correlation coefficient, To find the covariance, for time Sequence value, for time Sequence value, for standard deviation for standard deviation and All data are from earthquake frequency time series. To delay the lag order corresponding to the cross-correlation coefficient, For time; select earthquakes exceeding the specified time from the earthquake frequency time series. The maximum lag cross-correlation coefficient and its corresponding lag order for intervals of 2 standard deviations (95% confidence interval); Based on the earthquake frequency time series, the maximum delay cross-correlation coefficient, and their corresponding lag orders, a delay-related influence path network for the earthquake frequency time series is constructed. Then, based on different combinations of independent and dependent variables, a stepwise method is used to screen the independent variables, resulting in a regression model, which specifically includes: Using earthquake frequency time series as nodes, maximum delay cross-correlation coefficients as edges, and their corresponding lag orders as the direction and weight of the edges, a delay-related influence path network of earthquake frequency time series is constructed; the edges pointing from one earthquake frequency time series to another are marked as pointing from the independent variable series to the dependent variable series. Traverse the edges of the earthquake frequency time series delay correlation path network to obtain several paths, and generate a combination for each path with the lagged form of the independent variable sequence as the independent variable and the current value of the dependent variable sequence as the dependent variable, to obtain different combinations of independent and dependent variables. Based on different combinations of independent and dependent variables, the stepwise method is used to screen the independent variables and obtain the regression model.

[0028] The regression model with a lag order less than 0 is selected from the regression models, and the known independent variable data is input to output the predicted value of future earthquake frequency. That is, the actual value of earthquake frequency within a certain magnitude range of a certain fault zone is statistically analyzed. Using the lag effect of the independent variable, the value of the dependent variable with delayed correlation and its confidence interval are predicted and calculated, that is, the mean value of earthquake frequency and its confidence interval for a certain magnitude and time range of other fault zones. At the same time, according to the practical meaning of earthquake frequency and its confidence interval, that is, earthquake frequency is an integer and confidence interval is a non-negative interval, the predicted value of the dependent variable and its confidence interval are processed. If the dependent variable value contains a decimal, the result is rounded up. If a negative value or a confidence interval containing a negative value is obtained, the negative value or negative confidence upper and lower limits are corrected to 0. Finally, the predicted value of earthquake frequency and its confidence interval that conform to the practical meaning are obtained.

[0029] The beneficial effects of this invention are as follows: By constructing a delayed correlation path network of earthquake frequency time series using a one-way connected graph, and sorting out different combinations of independent and dependent variables that can be used to construct a regression model, the independent variables are screened using stepwise regression, thereby fitting a regression model in which the overall equation and regression coefficients pass the significance test and have weak multicollinearity. This invention achieves a quantitative and structured model expression of the delayed correlation between earthquake frequency time series, providing reliable and interpretable model support for earthquake frequency prediction based on statistical laws, and overcoming the shortcomings of existing methods in combining quantitative prediction with mechanism.

[0030] Example 2 This invention provides a schematic flowchart of an earthquake frequency prediction method based on delay correlation. Building upon Example 1, to make the invention more concrete and credible, it is applied to the known Western Sichuan Y-type fault system. Earthquake catalog data from 1970 to 2023 for the Longmenshan, Xianshuihe, Anninghe-Zemuhe, and Daliangshan fault zones of the Western Sichuan Y-type fault system are obtained. Each data entry includes information such as earthquake occurrence time, longitude, latitude, and magnitude. Fault zones are located and matched to this data, and the same method is used to determine the magnitude. A total of 5698 earthquake catalog data entries are obtained for analysis. An overview of the earthquake catalog data for each fault zone is shown in Table 1.

[0031] Table 1

[0032] The number of earthquakes of magnitude greater than or equal to each fault zone in the Western Sichuan Y-type fault system was statistically analyzed. The least squares method was used to fit the GR relationship of the target fault system, and the results are as follows: Figure 2 As shown, the horizontal axis represents the magnitude. The vertical axis represents the logarithmic fitted values ​​of earthquake frequencies. The line represents the logarithmic fitting curve of earthquake frequency, and the points represent the measured logarithmic values ​​of earthquake frequency. From Figure 2It can be seen that the measured values ​​of the Longmenshan fault zone M < 3.0 and the Anninghe-Zemuhe and Daliangshan fault zones M < 3.0 are... N represents the number of earthquakes of the corresponding magnitude and below, all of which are less than the fitted value. Therefore, it can be inferred that the earthquake events with M < 3.0 along these two fault zones are incomplete, and their Mc values ​​are both M 3.0; while the measured values ​​along the Xianshuihe fault zone... Then all are in the fitting The magnitude fluctuates within a small range, indicating that the collected earthquake events along the fault zone are complete, and Mc is the minimum complete magnitude M2.3 in its earthquake catalog. After removing earthquake events with magnitudes less than Mc from all fault zones, a complete earthquake catalog with magnitudes greater than or equal to Mc is obtained, as shown in Table 2.

[0033] Table 2

[0034] As shown in Table 2, the Mc of the Longmenshan Fault Zone and the Anninghe-Zemuhe and Daliangshan Fault Zone is M3.0, while the Mc of the Xianshuihe Fault Zone is M2.3. To ensure the completeness of the data and the comparability of the analysis, earthquake catalog data with M≥3.0 were used to construct earthquake frequency time series for each fault zone. In the Western Sichuan Y-type fault system, "year" is a suitable time unit for statistical earthquake frequency based on the amount of data. In addition, shallow-focus earthquakes with M≥4.5 are generally considered to be of high magnitude and destructive. Therefore, the statistical earthquake frequency is further divided into two groups: M=3.0~4.4 and M≥4.5, to study the differences and correlations between the lower magnitude group and the higher magnitude group. Thus, nine groups of earthquake frequency time series for each fault zone were obtained, each with a series length of 54 periods (1970~2023). These were coded for subsequent analysis and description, as shown in Table 3. Table 3

[0035] In addition, to present the coordinated changes in earthquake frequency across different fault zones, the minimum-maximum normalization method was used to transform the earthquake frequency time series of all fault zones in each magnitude group.

[0036] To quantitatively measure the delayed correlation (hysteresis correlation) of the mutual influence of earthquakes between different fault zones, and to characterize whether there is a delay and its magnitude in the mutual triggering or inhibition of earthquakes between different fault zones, the delay cross-correlation coefficients between the time series of each earthquake frequency were calculated. The maximum delay correlation coefficients exceeding 2 standard deviations are shown in Table 4.

[0037] Table 4

[0038] This indicates that some sequence pairs exhibit significant delayed cross-correlation, leading to important insights into the correlation of earthquake frequencies across different fault zones: ① In the complete magnitude group and low magnitude group, the Longmenshan fault zone lags behind the Xianshuihe fault zone by 14 or 15 years, but not by the Anninghe-Zemuhe and Daliangshan fault zones; ② In the high magnitude group, the Xianshuihe fault zone, Anninghe-Zemuhe, and Daliangshan fault zones lag behind the Longmenshan fault zone by 36 years and 6 years, respectively; ③ Compared to the low magnitude group of the Longmenshan fault zone, the Xianshuihe fault zone lags behind the Anninghe-Zemuhe and Daliangshan fault zones by 36 years and 6 years, respectively; The high-magnitude group of the Hehe Fault Zone lags behind by 36 years, and the high-magnitude group of the Anninghe-Zemuhe and Daliangshan Fault Zone lags behind by 6 years; ④ The high-magnitude groups of the Longmenshan Fault Zone and the Anninghe-Zemuhe and Daliangshan Fault Zone lag behind the low-magnitude group of the Xianshuihe Fault Zone by 14 years and 21 years, respectively; ⑤ The high-magnitude group of the Longmenshan Fault Zone has no lag relative to the low-magnitude group, and also no lag relative to the low-magnitude group of the Anninghe-Zemuhe and Daliangshan Fault Zone; ⑥ The high-magnitude group of the Anninghe-Zemuhe and Daliangshan Fault Zone lags behind the low-magnitude group by 6 years. The above correlations are illustrated in a schematic diagram, such as... Figure 3 The diagram shown is a network diagram illustrating the path of the correlation between earthquake frequency time series and their effects.

[0039] Based on the correlations and delays of the time series of earthquake frequencies obtained from the above analysis, a regression model is established to accurately describe the quantitative relationships between these correlations. Figure 3 We established a multiple linear regression model for two types of earthquake frequency time series: complete magnitude group, low magnitude group, and high magnitude group. In addition, we used unnormalized earthquake frequency time series to make the practical meaning of the regression equation more intuitive.

[0040] according to Figure 3 The dependent and independent variables of the regression equation can be represented by the variable combination form of the earthquake frequency time series regression model as shown in Table 5. Table 5

[0041] By using the stepwise method to screen the independent variables, the following model was obtained, in which both the equation and the regression coefficients of the independent variables are significant and the multicollinearity is weak:

[0042] The above formulas can be used to calculate the mean earthquake frequency and its confidence interval for the corresponding magnitude group of each fault zone. In actual interpretation, the results with decimals should be rounded up to conform to the actual meaning of earthquake frequencies being integers, and also to make the expectations reasonable. Among them, due to the lag effect of the independent variable, formulas (3) to (5) can be used to predict the dependent variable (M≥4.5 for the Anninghe-Zemuhe and Daliangshan fault zones, and M≥4.5 for the Xianshuihe fault zone). Although other equations cannot make predictions, they also reveal the quantitative correlation between the earthquake frequencies of each fault zone. The above formulas Between 45.9% and 100%, equation (5) represents a perfect fit, and the reliability of the model results, except for equation (3), exceeds 60%, indicating high credibility. In addition, when using equations (1), (5), and (8) for calculation, negative values ​​or confidence intervals containing negative values ​​may be obtained. Based on the practical significance of earthquake frequencies being integers, the upper and lower confidence limits for negative values ​​or negative values ​​should be corrected to 0.

[0043] The beneficial effects of this invention are as follows: By constructing a delayed correlation path network of earthquake frequency time series using a one-way connected graph, and sorting out different combinations of independent and dependent variables that can be used to construct a regression model, the independent variables are screened using stepwise regression, thereby fitting a regression model in which the overall equation and regression coefficients pass the significance test and have weak multicollinearity. This invention achieves a quantitative and structured model expression of the delayed correlation between earthquake frequency time series, providing reliable and interpretable model support for earthquake frequency prediction based on statistical laws, and overcoming the shortcomings of existing methods in combining quantitative prediction with mechanism.

Claims

1. A method for predicting earthquake frequency based on delay correlation, characterized in that, Includes the following steps: Obtain earthquake catalog data within the fault system region, assign all earthquake events to the corresponding fault zones in the fault system based on the earthquake catalog data, and count the frequency of earthquakes of magnitude greater than or equal to each fault zone to obtain a complete earthquake catalog. Based on the preset magnitude range and time unit, the earthquake frequency of each fault zone in the complete earthquake catalog is statistically analyzed and arranged in chronological order to obtain the earthquake frequency time series. Calculate the delay cross-correlation coefficient between each pair of data in the earthquake frequency time series, and select the maximum delay cross-correlation coefficient and its corresponding lag order from the earthquake frequency time series based on the delay cross-correlation coefficient; Based on the earthquake frequency time series, the maximum delay cross-correlation coefficient and its corresponding lag order, a delay-related influence path network of the earthquake frequency time series was constructed. Based on different combinations of independent and dependent variables, the stepwise method was used to screen the independent variables and obtain the regression model. Select regression models with lag order less than 0 from the regression models, input the known independent variable data, and output the predicted value of future earthquake frequency.

2. The earthquake frequency prediction method based on delay correlation according to claim 1, characterized in that, The obtained complete earthquake catalog specifically includes: Obtain earthquake catalog data within the fault system region, and assign all earthquake events to the corresponding fault zones in the fault system based on the earthquake locations in the earthquake catalog data; The frequency of earthquakes with magnitudes greater than or equal to each fault zone was statistically analyzed, and the Gutenberg-Richard relation was fitted using the least squares method to obtain the theoretical relationship between earthquake frequency and magnitude, i.e., the logarithmic fitted value of earthquake frequency. The logarithmic fitted values ​​of earthquake frequencies for each fault zone are compared with the measured logarithmic values ​​of earthquake frequencies. When the measured logarithmic values ​​of earthquake frequencies for each fault zone are all less than the logarithmic fitted values, the magnitude corresponding to the logarithmic fitted values ​​of earthquake frequencies is taken as the minimum complete magnitude. In the earthquake catalog data, all earthquakes with magnitudes smaller than the minimum complete magnitude are filtered out to obtain a complete earthquake catalog.

3. The earthquake frequency prediction method based on delay correlation according to claim 1, characterized in that, The expression for the logarithmic fitted value of the earthquake frequency is as follows: in, The logarithmic fitted value of earthquake frequency represents the theoretical relationship between earthquake frequency and magnitude. For a certain magnitude, For a certain magnitude Earthquake frequency of magnitudes 1 and below and These are all coefficients that need to be estimated. This represents the overall seismic activity within the statistical spatiotemporal range. This represents the activity of high-magnitude earthquakes within the spatiotemporal range of the statistics.

4. The earthquake frequency prediction method based on delay correlation according to claim 1, characterized in that, The obtained earthquake frequency time series specifically includes: Based on the goals and analysis requirements of earthquake prediction, the magnitude range and time unit of all fault zones are uniformly set to obtain the preset magnitude range and time unit. Based on the preset magnitude range and time unit, the number of earthquake events in each fault zone within the time unit is counted in the complete earthquake catalog to obtain the original earthquake frequency sequence of each fault zone. Based on the original earthquake frequency sequence, the earthquake frequency time series is obtained by arranging them in chronological order and normalizing them.

5. The earthquake frequency prediction method based on delay correlation according to claim 1, characterized in that, The expression for the delayed cross-correlation coefficient is as follows: in, To delay the cross-correlation coefficient, To find the covariance, for time Sequence value, for time Sequence value, for standard deviation for standard deviation and All data are from earthquake frequency time series. To delay the lag order corresponding to the cross-correlation coefficient, For time; select from the earthquake frequency time series that exceeds the specified time. The maximum delay cross-correlation coefficient and its corresponding lag order for the interval of 2 times the standard deviation.

6. The earthquake frequency prediction method based on delay correlation according to claim 1, characterized in that, The obtained regression model specifically includes: Using earthquake frequency time series as nodes, maximum delay cross-correlation coefficients as edges, and their corresponding lag orders as the direction and weight of the edges, a delay-related influence path network of earthquake frequency time series is constructed; the edges pointing from one earthquake frequency time series to another are marked as pointing from the independent variable series to the dependent variable series. Traverse the edges of the earthquake frequency time series delay correlation path network to obtain several paths, and generate a combination for each path with the lagged form of the independent variable sequence as the independent variable and the current value of the dependent variable sequence as the dependent variable, to obtain different combinations of independent and dependent variables. Based on different combinations of independent and dependent variables, the stepwise method is used to screen the independent variables and obtain the regression model.