Evaluation of earthquake precursor prediction effectiveness and prediction parameter optimization method

Abstract

The present invention provides a method for evaluating earthquake precursor prediction effectiveness and optimizing prediction parameters. The method preprocesses earthquake precursor signals and earthquake catalog data, performs a grid search on parameter combinations to be optimized to generate multiple different parameter combinations, uses a preset prediction strategy to predict the preprocessed data based on the multiple different parameter combinations, generates a Molchan plot corresponding to each parameter combination based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution, and determines the optimal parameter combination from the multiple different parameter combinations based on the first surface fraction value between the actual prediction curve and the confidence interval in each Molchan plot. The method can eliminate the impact of differences in the number of earthquake catalog samples on earthquake precursor prediction effectiveness evaluation, improve the optimization efficiency of multiple parameters, and enhance the statistical significance of earthquake precursor prediction effectiveness evaluation results.

Description

Technical Field

[0001] The present invention relates to the technical field of earthquake prediction and effectiveness evaluation, and in particular to a method for earthquake precursor prediction effectiveness evaluation and prediction parameter optimization. Background Art

[0002] Existing earthquake precursor prediction effectiveness evaluation technology mainly uses the method of drawing a Molchan diagram (the horizontal axis is the alarm rate, the vertical axis is the success rate) and calculating the area fraction value (i.e., the S value) between the actual curve and the diagonal line in the Molchan diagram. The limitations of this method are mainly as follows: it requires the same number of samples in different earthquake catalogs, making it difficult to compare earthquake events with different sample numbers due to different magnitudes, epicenter distances, and focal depths; parameters such as time, space, and intensity need to be optimized step by step, which is inefficient; there is a lack of dynamic confidence interval testing for prediction effectiveness, and statistical significance is insufficient. Summary of the Invention

[0003] In view of this, the object of the present invention is to provide a method for earthquake precursor prediction effectiveness evaluation and prediction parameter optimization to alleviate the above-mentioned problems existing in the existing earthquake precursor prediction effectiveness evaluation technology.

[0004] In the first aspect, an embodiment of the present invention provides a method for evaluating the effectiveness of earthquake precursor prediction and optimizing prediction parameters, comprising: preprocessing earthquake precursor signals and earthquake catalog data, and performing grid search on parameter combinations to be optimized to generate multiple different parameter combinations; wherein the parameters to be optimized in the parameter combinations to be optimized include the lower limit of magnitude, epicenter distance, focal depth, lead time and forecast window length, and the number of earthquake catalog samples corresponding to different parameter combinations is different; using a preset prediction strategy to predict the preprocessed data based on multiple different parameter combinations to obtain prediction results corresponding to each parameter combination; wherein the prediction results include the time point of occurrence of earthquake precursor abnormal events and / or the time of occurrence of earthquake events within a preset time period. point; based on the prediction results corresponding to each parameter combination and the number of earthquake catalog samples and the preset probability distribution, a Molchan diagram corresponding to each parameter combination is generated; wherein the probability distribution represents the probability of successfully predicting different numbers of earthquakes under different alarm rates, and each Molchan diagram includes a corresponding actual prediction curve and a confidence interval, and the actual prediction curve represents a first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination, and the confidence levels of the confidence intervals corresponding to different Molchan diagrams are the same; determine a first area fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, and determine the optimal parameter combination from a plurality of different parameter combinations based on the first area fraction value.

[0005] In the second aspect, an embodiment of the present invention further provides an earthquake precursor prediction effectiveness evaluation and prediction parameter optimization device, comprising: a processing module for preprocessing earthquake precursor signals and earthquake catalog data, and performing a grid search on the parameter combination to be optimized to generate a plurality of different parameter combinations; wherein the parameters to be optimized in the parameter combination to be optimized include the lower limit of magnitude, epicenter distance, focal depth, lead time and forecast window length, and the number of earthquake catalog samples corresponding to different parameter combinations is different; a prediction module for using a preset prediction strategy to predict the preprocessed data based on a plurality of different parameter combinations to obtain the prediction results corresponding to each parameter combination; wherein the prediction results include the time point of occurrence of the earthquake precursor abnormal event and / or the time of occurrence of the earthquake event within a preset time period point; a generation module, used to generate a Molchan diagram corresponding to each parameter combination based on the prediction results corresponding to each parameter combination and the number of earthquake catalog samples and a preset probability distribution; wherein the probability distribution represents the probability of successfully predicting different numbers of earthquakes under different alarm rates, and each Molchan diagram includes a corresponding actual prediction curve and a confidence interval, and the actual prediction curve represents a first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination, and the confidence levels of the confidence intervals corresponding to different Molchan diagrams are the same; a determination module, used to determine a first area fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, and determine the optimal parameter combination from multiple different parameter combinations based on the first area fraction value.

[0006] In a third aspect, an embodiment of the present invention further provides an electronic device comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method described in the first aspect above.

[0007] An embodiment of the present invention provides a method for evaluating the effectiveness of earthquake precursor prediction and optimizing prediction parameters. First, earthquake precursor signals and earthquake catalog data are preprocessed, and a grid search is performed on parameter combinations to be optimized (including the lower limit of magnitude, epicentral distance, focal depth, lead time, and prediction window length) to generate multiple different parameter combinations. Then, a preset prediction strategy is used to predict the preprocessed data based on the multiple different parameter combinations, and a Molchan diagram corresponding to each parameter combination is generated based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution. Finally, the optimal parameter combination is determined from the multiple different parameter combinations based on the first surface fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, so that the optimal parameter combination can be subsequently applied to perform earthquake precursor prediction. By using the above technology, earthquake precursor signals and earthquake catalog data as well as multiple different parameter combinations obtained through grid search can be used for prediction to generate multiple Molchan diagrams. Therefore, earthquake precursor prediction performance evaluation and parameter combination optimization can be achieved by comparing the first surface fraction values ​​between the actual prediction curves and the confidence intervals in different Molchan diagrams. Since different parameter combinations correspond to different numbers of earthquake catalog samples, the influence of the difference in the number of earthquake catalog samples on the earthquake precursor prediction performance evaluation can be eliminated, and the earthquake precursor prediction performance evaluation of any earthquake catalog can be supported. Since multiple parameters are optimized in the form of parameter combinations, the lower limit of magnitude, epicentral distance, focal depth, lead time and prediction window length can be optimized simultaneously, thereby improving the optimization efficiency of multiple parameters. Since confidence intervals with a certain confidence level are generated in combination with probability distribution when generating Molchan diagrams, dynamic confidence interval testing of earthquake precursor prediction performance can be achieved, thereby improving the statistical significance of earthquake precursor prediction performance evaluation results, thereby facilitating the improvement of the accuracy of short-term earthquake prediction.

[0008] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0009] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0011] Figure 1 Schematic diagram of a flow chart of a method for earthquake precursor prediction effectiveness evaluation and prediction parameter optimization in an embodiment of the present invention;

[0012] Figure 2 Schematic diagram of a prediction strategy based on precursor anomalies in an embodiment of the present invention;

[0013] Figure 3 Schematic diagram of the S value between the actual prediction curve and the random line in an embodiment of the present invention;

[0014] Figure 4 Schematic diagram of the 95% confidence interval of the Molchan diagram corresponding to different numbers of earthquake samples in an embodiment of the present invention;

[0015] Figure 5 The difference between the actual prediction curve and the 95% confidence interval in the embodiment of the present invention is Value diagram;

[0016] Figure 6 This is an overall flow chart of the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method in an embodiment of the present invention;

[0017] Figure 7 The electromagnetic anomaly of the actual KAK station in the embodiment of the present invention is the optimal Schematic diagram of the Molchan error diagram when the value is;

[0018] Figure 8 Schematic diagram of the structure of an earthquake precursor prediction effectiveness evaluation and prediction parameter optimization device in an embodiment of the present invention;

[0019] Figure 9 Schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] At present, the existing earthquake precursor prediction effectiveness evaluation technology uses the method of drawing a Molchan diagram (the horizontal axis is the alarm rate, the vertical axis is the success rate) and calculating the area fraction value (i.e., the S value) between the actual curve and the diagonal line in the Molchan diagram. The main limitations are as follows: the number of samples in different earthquake catalogs must be consistent, which makes it difficult to compare earthquake events with different magnitudes, epicenter distances, and focal depths that result in different sample numbers; parameters such as time, space, and intensity need to be optimized step by step, which is inefficient; there is a lack of dynamic confidence interval testing for prediction effectiveness, and statistical significance is insufficient.

[0022] Based on this, the present invention provides a method for evaluating the effectiveness of earthquake precursor prediction and optimizing prediction parameters, which can alleviate the above-mentioned problems existing in the existing earthquake precursor prediction effectiveness evaluation technology.

[0023] To facilitate understanding of this embodiment, firstly, a method for evaluating earthquake precursor prediction performance and optimizing prediction parameters disclosed in an embodiment of the present invention is described in detail. Figure 1 As shown, the method may include the following steps:

[0024] Step S102 : pre-processing the earthquake precursor signal and earthquake catalog data, and performing a grid search on the parameter combination to be optimized to generate a plurality of different parameter combinations.

[0025] Among them, the parameters to be optimized in the parameter combination to be optimized may include the lower limit of magnitude (M), epicenter distance (R), focal depth (D), lead time ( ), prediction window length (L), etc. Different parameter combinations correspond to different numbers of earthquake catalog samples.

[0026] Step S104 , predicting the preprocessed data based on a plurality of different parameter combinations using a preset prediction strategy, and obtaining prediction results corresponding to each parameter combination.

[0027] The prediction results may include the time point at which an abnormal earthquake precursor event occurs and / or the time point at which an earthquake event occurs within a preset time period.

[0028] Step S106: Generate a Molchan diagram corresponding to each parameter combination based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution.

[0029] Among them, the probability distribution can represent the probability of successfully predicting different numbers of earthquakes under different alarm rates. Each Molchan diagram can include a corresponding actual prediction curve and confidence interval. The actual prediction curve can represent the first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination. The confidence levels of the confidence intervals corresponding to different Molchan diagrams are the same.

[0030] Step S108: determining a first surface fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, and determining an optimal parameter combination from a plurality of different parameter combinations based on the first surface fraction value.

[0031] An embodiment of the present invention provides a method for evaluating the effectiveness of earthquake precursor prediction and optimizing prediction parameters. First, earthquake precursor signals and earthquake catalog data are preprocessed, and a grid search is performed on parameter combinations to be optimized (including the lower limit of magnitude, epicentral distance, focal depth, lead time, and prediction window length) to generate multiple different parameter combinations. Then, a preset prediction strategy is used to predict the preprocessed data based on the multiple different parameter combinations, and a Molchan diagram corresponding to each parameter combination is generated based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution. Finally, the optimal parameter combination is determined from the multiple different parameter combinations based on the first surface fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, so that the optimal parameter combination can be subsequently applied to perform earthquake precursor prediction. By using the above technology, earthquake precursor signals and earthquake catalog data as well as multiple different parameter combinations obtained through grid search can be used for prediction to generate multiple Molchan diagrams. Therefore, earthquake precursor prediction performance evaluation and parameter combination optimization can be achieved by comparing the first surface fraction values ​​between the actual prediction curves and the confidence intervals in different Molchan diagrams. Since different parameter combinations correspond to different numbers of earthquake catalog samples, the influence of the difference in the number of earthquake catalog samples on the earthquake precursor prediction performance evaluation can be eliminated, and the earthquake precursor prediction performance evaluation of any earthquake catalog can be supported. Since multiple parameters are optimized in the form of parameter combinations, the lower limit of magnitude, epicentral distance, focal depth, lead time and prediction window length can be optimized simultaneously, thereby improving the optimization efficiency of multiple parameters. Since confidence intervals with a certain confidence level are generated in combination with probability distribution when generating Molchan diagrams, dynamic confidence interval testing of earthquake precursor prediction performance can be achieved, thereby improving the statistical significance of earthquake precursor prediction performance evaluation results, thereby facilitating the improvement of the accuracy of short-term earthquake prediction.

[0032] As a possible implementation, the grid search of the parameter combination to be optimized to generate multiple different parameter combinations in the above-mentioned step S102 may include: constructing a parameter grid of the parameter combination to be optimized based on the search range and search step of each parameter to be optimized; wherein the parameter combination is represented by grid points in the parameter grid; and traversing the parameter grid to extract the parameter combination represented by each grid point in the parameter grid to obtain multiple different parameter combinations.

[0033] In actual earthquake magnitude lower limit (M), epicenter distance (R), focal depth (D), lead time ( ), forecast window length (L) for grid search, you can first set M, R, D, , L's respective search strategies (including search range and search step), for example, set the search strategy of M to be in step 0.1 Search within the range of , set the search strategy of R to be in the range of 10km Search within the range of , set the search strategy of D to be in the range of 10km Search within the range, set The search strategies of L and L are to search within the range of 1 to 40 days with a step size of 1 day. Then, the parameter grid is divided according to the search strategy set for each parameter. Each grid point in the parameter grid represents a corresponding set of parameter values ​​(including M value, R value, D value, value and L value), different grid points represent different groups of parameter values; then the grid points of the parameter grid are traversed to obtain the corresponding group of parameter values ​​represented by each grid point, and multiple groups of parameter values ​​are obtained, that is, after the value of each parameter changes, a new group of parameter values ​​can be formed with the current values ​​of the other parameters. The number of earthquake catalog samples corresponding to different groups of parameter values ​​is different.

[0034] As a possible implementation method, the pre-processed data can be a time series that characterizes the change of abnormal index values ​​over time within a preset time period; wherein, the abnormal index can be defined according to actual applications, and specifically signal energy or other indicators can be selected as abnormal indicators. The specific definition method of the abnormal index is not limited here. Based on this, the preset prediction strategy adopted in the above step S104 may include: for each parameter combination, traversing the preset abnormal threshold set to use each abnormal threshold in the abnormal threshold set to determine whether there is an earthquake precursor abnormal signal in the time series whose abnormal index value exceeds the corresponding abnormal threshold, and determining the time point corresponding to the earthquake precursor abnormal signal as the time point of the earthquake precursor abnormal event, and then determining the forecast time window within the preset time period based on the determined earthquake precursor abnormal event occurrence time point and the lead time and forecast window length in the parameter combination, and judging whether there is an earthquake event occurrence time point within each forecast time window based on the magnitude lower limit, epicenter distance and focal depth in the parameter combination.

[0035] Before actually predicting earthquake precursors, a prediction model based on earthquake precursors can be established, and corresponding prediction strategies can be set for the prediction model based on earthquake precursor anomalies. Figure 2 As shown, the energy of the residual signal is used as an abnormality indicator. The prediction strategy based on the earthquake precursor anomaly setting is as follows: In each prediction, the lead time ( ) and forecast window length (L) to describe the warning parameters, The units of and L are both days. For a time series where the residual signal energy value changes with the number of days (such as Figure 2 For a time series with time as the horizontal axis and residual signal energy value as the vertical axis, each day corresponds to a corresponding energy value. Figure 2 The days) exceeds a given threshold (e.g. Figure 2 After the abnormal threshold in is set to 1.5), Day to day Days (the time range is a time window with a window length of L) an alarm is issued, and when the residual energy exceeds a given threshold, an earthquake precursor abnormal event occurs. The time point when the residual energy exceeds the given threshold (such as Figure 2 The The time point of the earthquake precursor abnormal event is the time point of the alarm, and the period of time when the alarm is issued is the forecast time window. It is necessary to predict the time point of the earthquake event when the magnitude lower limit M, epicenter distance R and focal depth D meet specific conditions (determined by the values ​​of M, R and D in the specific parameter group) (such as Figure 2 The time point corresponding to the predicted earthquake is the time point in the forecast window). Only earthquake events that occur within the forecast time window can be predicted (such as Figure 2 The time point of occurrence of earthquakes predicted in the forecast time window) cannot be predicted, but earthquake events occurring outside the forecast time window (such as Figure 2 The time point of occurrence of the earthquake predicted by the leakage.

[0036] After setting up a prediction strategy for the prediction model based on earthquake precursor anomalies, all anomaly thresholds can be traversed to predict the time series under the corresponding anomaly threshold using each anomaly threshold according to the set prediction strategy, thereby obtaining the predicted time points of earthquake precursor anomaly events and / or earthquake events under different anomaly thresholds, so that the prediction results under different anomaly thresholds can be used to generate Molchan diagrams and then use Molchan diagrams to quantitatively evaluate the earthquake precursor prediction performance of the prediction model. Figure 2 For example, the anomaly thresholds can be 0.5, 1, 1.5, 2 and 2.5. The time series can be predicted at different anomaly thresholds according to the above-mentioned prediction strategy, and the prediction results at different anomaly thresholds can be obtained. The prediction results at each anomaly threshold include the corresponding time point of occurrence of the earthquake precursor anomaly event and / or the time point of occurrence of the earthquake event.

[0037] As a possible implementation, the above step S106 (i.e., generating a Molchan diagram corresponding to each parameter combination based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution) may include:

[0038] Step A1: Based on the prediction results corresponding to each parameter combination, generate the actual prediction curve corresponding to each parameter combination.

[0039] The actual prediction curve uses the alarm rate as the independent variable and the success rate as the dependent variable; based on this, the operation method of the above-mentioned step A1 may include: for each parameter combination, determining the alarm rate and success rate of the parameter combination corresponding to each abnormal threshold value based on the time point of occurrence of the earthquake precursor abnormal event and / or the time point of occurrence of the earthquake event corresponding to the parameter combination, and generating the actual prediction curve corresponding to the parameter combination based on the alarm rate and success rate corresponding to the parameter combination.

[0040] Continuing from the previous example, for the time series to be predicted, we can set is the total number of days in the time series to be predicted, is the number of days with earthquakes greater than a given magnitude, To predict the number of days with earthquakes based on geomagnetic anomalies, The number of days when an earthquake is predicted and actually occurs is: The number of days when earthquakes are predicted but there are no earthquakes is The number of days when earthquakes were actually detected despite predictions of no earthquakes; the alarm rate is the ratio of total forecast days to total days, that is, Success rate is the ratio of successful forecast days to the total forecast days, that is, The introduction and detailed description of the relevant variables are shown in Table 1.

[0041] Table 1 Schematic diagram of predicted earthquake / no earthquake and actual earthquake / no earthquake

[0042]

[0043] For each set of parameter values, after using the set parameter values ​​and each anomaly threshold to predict the time series under the corresponding anomaly threshold according to the set prediction strategy, the prediction results corresponding to the set of parameter values ​​at different anomaly thresholds (including the predicted time points of earthquake precursor anomaly events and / or the time points of earthquake events) can be used to calculate the alarm rate and success rate corresponding to the different anomaly thresholds. Then, the actual prediction curve is drawn on the Molchan diagram with the alarm rate as the independent variable and the success rate as the dependent variable. When the alarm rate is 0, no alarm is issued, and the success rate is also 0; the success rate increases as the alarm rate increases. When the alarm rate is 1, that is, alarms are issued on all days, the success rate is also 1.

[0044] Step A2: Generate a confidence interval corresponding to each parameter combination based on the number of earthquake catalog samples and the probability distribution corresponding to each parameter combination.

[0045] Exemplarily, the probability distribution may adopt a binomial distribution; based on this, the operation method of the above-mentioned step A2 may be: based on the number of earthquake catalog samples corresponding to each parameter combination and the binomial distribution, determine the range of the number of earthquakes successfully predicted for each parameter combination under different alarm rates; wherein the sum of the probability sizes of the range of earthquake numbers corresponding to each alarm rate matches the confidence level of the corresponding confidence interval; and then determine the confidence interval corresponding to each parameter combination based on the number of earthquake catalog samples and the range of earthquake numbers corresponding to each parameter combination.

[0046] In step A3, the actual prediction curve and confidence interval corresponding to each parameter combination are combined into a corresponding Molchan diagram to obtain the Molchan diagram corresponding to each parameter combination.

[0047] As a possible implementation, the confidence level may be 95%. Based on this, the calculation formula for the first area fraction value may be:

[0048]

[0049] in, is the first surface integral value, For the success rate, is the upper limit of the confidence interval, is the alarm rate.

[0050] As a possible implementation method, the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method may further include:

[0051] Step a1: adding a reference line to the target Molchan diagram corresponding to the optimal parameter combination based on the preset second relationship between the alarm rate and the success rate.

[0052] Step a2: determine the second area fraction value between the actual prediction curve and the reference line in the target Molchan diagram, and evaluate the earthquake precursor prediction effectiveness corresponding to the optimal parameter combination based on the first area fraction value and the second area fraction value corresponding to the target Molchan diagram.

[0053] Continuing with the previous example, the horizontal axis of the Molchan diagram is the alarm rate, and the vertical axis of the Molchan diagram is the success rate. In random situations, the alarm rate and the success rate are generally equivalent. In order to evaluate the effectiveness of the established prediction model in predicting earthquakes based on earthquake precursor abnormal signals relative to random situations, a diagonal line can be drawn on each obtained Molchan diagram as a reference line to represent the earthquake precursor prediction results under random situations; for each Molchan diagram, if the actual prediction curve corresponding to the prediction model in the Molchan diagram is above the diagonal line of the Molchan diagram, it can be considered that the earthquake precursor prediction effect of the prediction model is better than the earthquake precursor prediction effect under random situations; if the actual prediction curve corresponding to the prediction model in the Molchan diagram is below the diagonal line of the Molchan diagram, it can be considered that the earthquake precursor prediction effect of the prediction model is worse than the earthquake precursor prediction effect under random situations.

[0054] The earthquake precursor prediction performance of the prediction model under the corresponding parameter combination can be evaluated by analyzing the area fraction value between the actual prediction curve and the diagonal line in the Molchan diagram. Figure 3 As shown, the area fraction of the area enclosed by the actual prediction curve and the diagonal line in the Molchan diagram is set to S. The area enclosed by the actual prediction curve and the diagonal line is the part where the earthquake precursor prediction effect of the prediction model is better than that of the random case ( Figure 3 The area fraction of the actual prediction curve above the diagonal line is counted as a positive value, and the prediction effect of the earthquake precursor prediction model is worse than that of the random case ( Figure 3 The area fraction of the actual predicted curve (the blue area where the actual predicted curve is lower than the diagonal line) is counted as a negative value, then the area fraction S of the area enclosed by the actual predicted curve and the diagonal line in the Molchan diagram is the sum of the area fractions of all areas enclosed by the actual curve and the diagonal line. If S is greater than 0, it means that the actual predicted effect is better than the random prediction effect as a whole. If S is less than 0, it means that the actual predicted effect is worse than the random prediction effect as a whole. Figure 3 Taking the Molchan diagram shown in Figure 1 as an example, the S value in this Molchan diagram is -0.0489, indicating that the prediction model's actual prediction performance is generally worse than that of random prediction under the parameter combination corresponding to this Molchan diagram. Based on existing earthquake precursor anomaly data (i.e., the magnitude lower limit, epicentral distance, and focal depth are all fixed), different actual prediction curves can be generated by changing the lead time and / or the forecast window length, and thus different Molchan diagrams are obtained. Each Molchan diagram is quantified by calculating its S value to reflect the overall performance of the actual prediction compared to the random prediction.

[0055] In practical applications, after observing an abnormal earthquake precursor signal at a station, one usually wants to obtain the optimal spatiotemporal distribution of the earthquake catalog corresponding to the abnormal earthquake precursor signal (including the temporal distribution and spatial distribution of the earthquake catalog). For example, one wants to determine the earthquake with a large epicenter distance range and a large focal depth associated with the abnormal earthquake precursor signal. Therefore, it is necessary to establish the relationship between the abnormal information and the indicated earthquake in time, space and energy intensity, optimize the relevant parameters, and optimize the prediction parameters of the earthquake precursor signal. Since effective earthquake prediction requires information such as time, location and magnitude, the time information required for earthquake prediction here includes the lead time and the forecast window length, so it is necessary to develop a method that can optimize the relevant parameters at the same time. In the Molchan diagram, for a given alarm rate ,common earthquake samples, successfully predicted using random prediction methods The probability of an earthquake satisfies the following binomial distribution:

[0056]

[0057] in, represents the probability of the binomial distribution, is the total number of earthquakes, is the number of earthquakes successfully predicted, is the alarm rate; since the success rate is consistent with the alarm rate in the case of random guessing, the binomial distribution can be used to calculate the number of successful earthquakes predicted under different alarm rates. The probability of the total number of earthquakes is calculated Under the condition of , the probability of successfully predicting the number of earthquakes or less is 95% (i.e. How much is needed to make the probability in the binomial distribution not less than 95%), so as to calculate the probability of random prediction in the total number of earthquakes. The 95% confidence interval under this condition can be used to draw the 95% confidence interval range of the Molchan diagram corresponding to different earthquake sample numbers (such as Figure 4 shown).

[0058] Based on this, we can calculate the 95% confidence intervals for the random prediction method corresponding to different earthquake sample numbers. When the prediction curve exceeds the corresponding 95% confidence interval, it can be considered that the earthquake precursor anomaly signal contains significant predictive information. Figure 4 Shows different numbers of earthquake samples The corresponding 95% confidence interval shows that the confidence intervals corresponding to different earthquake sample numbers are quite different. The area fraction S is only applicable to optimizing the lead time when the earthquake sample is fixed (for example, given the magnitude lower limit M, epicenter distance R and focal depth D). and the forecast window length L, but there are limitations when comparing different earthquake samples (for example, the magnitude lower limit M, epicenter distance R, and focal depth D each vary within the corresponding range); in addition, when issuing warnings after observing earthquake precursor anomalies, different lead times are used. Different forecast window lengths L will affect the final forecast effect; therefore, a new area fraction can be introduced based on the area fraction S (abbreviated as ), this new area fraction can be used to compare earthquake precursor signals at different lead times. , the prediction performance of the earthquake catalog when the number of earthquake samples changes due to changes in relevant parameters (such as changes in the lower limit of magnitude M, epicenter distance R, and focal depth D) at different prediction window lengths L.

[0059] In actual application, for the Molchan diagram corresponding to each parameter combination, given a certain alarm rate, when the actual prediction curve is above the 95% confidence interval, it means that the actual prediction efficiency is significantly better than the random prediction efficiency; when the actual prediction curve is below the 95% confidence interval, it means that the actual prediction efficiency is significantly worse than the random prediction efficiency. The value can be calculated by calculating the algebraic sum of the area fractions of the area enclosed by the actual prediction curve and the 95% confidence interval in the Molchan diagram as value, A value greater than 0 indicates that the overall actual prediction performance is significantly better than the random prediction performance. A value less than 0 indicates that the overall actual prediction performance is not significantly better (or significantly worse) than the random prediction performance.

[0060] In actual application, the S value and the target Molchan diagram corresponding to the optimal parameter combination can also be calculated simultaneously. value, and then through the S value and The positive and negative values ​​and their sizes are used to evaluate the actual prediction performance under the optimal parameter combination.

[0061] As a possible implementation, determining the first area fraction value between the actual prediction curve and the confidence interval in each Molchan diagram in the above step S108 may include: determining the third area fraction value between the actual prediction curve and the horizontal axis in each Molchan diagram, and determining the fourth area fraction value between the confidence interval and the horizontal axis in each Molchan diagram; for each Molchan diagram, subtracting the fourth area fraction value corresponding to the Molchan diagram from the third area fraction value corresponding to the Molchan diagram to obtain the first area fraction value corresponding to the Molchan diagram.

[0062] In practical applications, in order to facilitate calculation, The value can also be calculated in the following way: For a Molchan diagram, first calculate the area of ​​the area enclosed by the actual prediction curve and the horizontal axis and the area enclosed by the 95% confidence interval and the horizontal axis, and then subtract the values ​​of the two calculated areas to get value.

[0063] In actual application, the implementation of the above earthquake precursor prediction performance evaluation and prediction parameter optimization method can mainly include the following steps:

[0064] Step 1: Establish a prediction model based on earthquake precursors and set the prediction strategy of the prediction model.

[0065] Step 2: Generate different parameter combinations (including M, R, D, , L) corresponding prediction curve, calculate the prediction curve under each parameter combination The value is used as the prediction performance evaluation indicator.

[0066] The value calculation formula is:

[0067]

[0068] like Figure 5 As shown, when the actual prediction curve is above the 95% confidence interval, the area enclosed by the actual prediction curve and the 95% confidence interval ( Figure 5 The area fraction of the actual prediction curve above the 95% confidence interval is positive. When the actual prediction curve is below the 95% confidence interval, the area fraction enclosed by the actual prediction curve and the 95% confidence interval is positive. Figure 5 The area fraction of the actual prediction curve (the blue area where the actual prediction curve is lower than the 95% confidence interval) is negative, and the algebraic sum of the area fractions of the entire area enclosed by the actual prediction curve and the 95% confidence interval can be calculated as value, A value greater than 0 means that the actual prediction performance is significantly better than the random prediction performance. A value less than 0 means that the actual prediction performance is not significantly better (or significantly worse) than the random prediction performance. The value of quantitatively examining the prediction performance of earthquake precursor prediction models for earthquake samples with different time information, different spatial information, and different intensity information, in order to optimize the lead time , prediction window length L, earthquake range (including epicenter distance R, focal depth D), magnitude lower limit M and other parameters to clarify the relationship between the precursory anomaly and the indicated earthquake in terms of time, space and intensity.

[0069] Since we can calculate and compare the The prediction performance of earthquake catalogs with different sample numbers can be compared by using the value of The value is used as a prediction performance evaluation indicator to eliminate the impact of the difference in the number of earthquake samples on the prediction performance evaluation, thereby realizing the prediction performance evaluation that eliminates the difference in the number of earthquake samples.

[0070] Step 3: Filter out The parameter combination with the largest value is taken as the optimal parameter combination, so that the parameters contained in the optimal parameter combination can be used later (advance time , prediction window length L, magnitude lower limit M, epicenter distance R and depth D) are used as earthquake prediction parameters in terms of time, space and intensity.

[0071] Compared with the existing technology, the core innovation of the above earthquake precursor prediction performance evaluation and prediction parameter optimization method lies in: The method of calculating the value of the earthquake prediction coefficient is used to eliminate the difference in the number of earthquake samples, and support the comparison of the prediction efficiency of any earthquake catalog; by introducing the parameter group optimization mechanism, the lower limit of magnitude, epicenter distance, focal depth, lead time and forecast window length are optimized simultaneously; by combining the binomial distribution to calculate the 95% confidence interval to quantitatively verify the significance of the prediction efficiency, a dynamic confidence test of the prediction efficiency is realized.

[0072] For ease of understanding, the implementation of the above-mentioned earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method is described below by taking a specific application as an example.

[0073] See also Figure 6 As shown, the earthquake precursor prediction performance evaluation and prediction parameter optimization method can be carried out according to the following steps:

[0074] Step S1: data preprocessing and Molchan graph generation.

[0075] First, input earthquake precursor abnormal signals (such as electromagnetic abnormal signals) and earthquake catalog data, and perform data preprocessing; then set the magnitude M, epicenter distance R, focal depth D, and lead time. , the forecast window length L, their respective search ranges and search steps, for M, R, D, , L is gridded to construct a parameter grid; then, the parameter grid is traversed to perform parameter grid search to generate multiple different parameter combinations, and a prediction curve is generated for each parameter combination and the confidence interval is calculated to obtain the Molchan diagram (including the prediction curve and the confidence interval) corresponding to each parameter combination. The specific implementation method can be found in the relevant content above and will not be repeated here.

[0076] Step S2, Value calculation and optimal parameter screening.

[0077] The Molchan diagram corresponding to each parameter combination can be Calculate the value and filter out the best value from multiple different parameter combinations The parameter combination with the largest value is regarded as the optimal parameter combination, and the parameters contained in the optimal parameter combination (i.e. M, R, D, , L) as the optimal parameters output, so that the optimal parameters can be applied to the earthquake early warning system later.

[0078] Taking KAK station as an example, we can input earthquake catalogs and electromagnetic anomaly data of magnitude 4 or above around KAK station in the past 10 years. First, we can get the following data based on the earthquake catalogs and the occurrence time of electromagnetic anomalies: Figure 2 The time series of earthquake or earthquake precursor anomalies is marked as shown; then the three parameters of earthquake in time, space and intensity (including M, R, D, Different earthquake catalogs are obtained by different combinations of L and L, and the 95% confidence interval of the binomial distribution under the number of earthquake samples corresponding to different earthquake catalogs is determined; for each earthquake catalog, different lead times and different forecast windows are traversed to generate actual prediction curves of different Molchan diagrams, and the prediction accuracy of the earthquake catalog based on earthquake precursor anomalies is calculated. value; finally, the search finds The earthquake time and space intensity parameters corresponding to the maximum value (including M, R, D, , L).

[0079] Figure 7 The actual electromagnetic anomaly of KAK station is shown in the optimal Value (i.e. The Molchan error diagram at the maximum value is shown in the Molchan error diagram. The value is 0.0714, the actual prediction curve is significantly higher than the 95% confidence interval, and the actual prediction curve is above the diagonal line.

[0080] because The calculation method of the value takes into account the confidence interval of a certain confidence level, so it can be directly compared The earthquake precursor anomaly prediction performance comparison of different earthquake samples is realized by using the value of Figure 7 For example, by traversing the parameters, the output is The optimal parameters corresponding to the maximum value are: 、 、 、 、 .

[0081] The beneficial effects of the above-mentioned earthquake precursor prediction performance evaluation and prediction parameter optimization method can be mainly reflected in the following aspects:

[0082] 1) Eliminate the limit on the number of earthquake samples: Introduced The calculation method can effectively eliminate the influence of the difference in the number of earthquake samples and support the comparison of the prediction performance of any earthquake catalog.

[0083] 2) Synchronous optimization of the three elements of time and space: achieving the lower limit of magnitude M, epicenter distance R, focal depth D, and lead time The integrated parameter optimization of the forecast window length L can avoid efficiency loss and error accumulation compared with the traditional step-by-step parameter optimization method.

[0084] 2) Dynamic confidence test mechanism: Combined with the binomial distribution to generate a 95% confidence interval, quantitatively verifying the statistical significance of the predictive performance.

[0085] 3) Parameter screening capability: Parameter optimization is performed using a grid search method, which can globally traverse and search for the optimal parameter combination.

[0086] 4) Wide applicability: The system is applicable to the prediction performance evaluation of various precursor signals (such as geomagnetic, geoelectric, geothermal, atmospheric, and ionospheric anomalies), and supports the optimization of earthquake prediction parameters in different regions and magnitude ranges.

[0087] 5) High industrial applicability: It can be directly integrated into the earthquake early warning system to optimize the configuration of earthquake precursor prediction parameters, improve the accuracy of short-term earthquake prediction, and provide a scientific basis for disaster prevention and mitigation.

[0088] Based on the above earthquake precursor prediction performance evaluation and prediction parameter optimization method, the embodiment of the present invention also provides an earthquake precursor prediction performance evaluation and prediction parameter optimization device, see Figure 8 As shown, the device may include the following modules:

[0089] Processing module 802 is used to preprocess earthquake precursor signals and earthquake catalog data, and perform grid search on the parameter combination to be optimized to generate multiple different parameter combinations; wherein the parameters to be optimized in the parameter combination to be optimized include the lower limit of magnitude, epicenter distance, focal depth, lead time and forecast window length, and the number of earthquake catalog samples corresponding to different parameter combinations is different.

[0090] The prediction module 804 is used to predict the preprocessed data based on multiple different parameter combinations using a preset prediction strategy to obtain prediction results corresponding to each parameter combination; wherein the prediction results include the time point of occurrence of earthquake precursor abnormal events and / or the time point of occurrence of earthquake events within a preset time period.

[0091] Generation module 806 is used to generate a Molchan diagram corresponding to each parameter combination based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples and the preset probability distribution; wherein the probability distribution represents the probability of successfully predicting different numbers of earthquakes under different alarm rates, and each Molchan diagram includes a corresponding actual prediction curve and a confidence interval, and the actual prediction curve represents the first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination, and the confidence levels of the confidence intervals corresponding to different Molchan diagrams are the same.

[0092] The determination module 808 is configured to determine a first surface fraction value between the actual prediction curve and the confidence interval in each Molchan diagram, and determine an optimal parameter combination from a plurality of different parameter combinations based on the first surface fraction value.

[0093] The above-mentioned earthquake precursor prediction effectiveness evaluation and prediction parameter optimization device can be used to make predictions using earthquake precursor signals and earthquake catalog data as well as multiple different parameter combinations obtained through grid search, thereby generating multiple Molchan diagrams. Thus, earthquake precursor prediction effectiveness evaluation and parameter combination optimization are achieved by comparing the first surface fraction values ​​between the actual prediction curves and the confidence intervals in different Molchan diagrams. Since different parameter combinations correspond to different numbers of earthquake catalog samples, the impact of the difference in the number of earthquake catalog samples on the earthquake precursor prediction effectiveness evaluation can be eliminated, and the earthquake precursor prediction effectiveness evaluation of any earthquake catalog can be supported. Since multiple parameters are optimized in the form of parameter combinations, the lower limit of magnitude, epicentral distance, focal depth, lead time and forecast window length can be optimized simultaneously, thereby improving the optimization efficiency of multiple parameters. Since confidence intervals with a certain confidence level are generated in combination with probability distribution when generating Molchan diagrams, dynamic confidence interval testing of earthquake precursor prediction effectiveness can be achieved, thereby improving the statistical significance of the earthquake precursor prediction effectiveness evaluation results, thereby facilitating the improvement of the accuracy of short-term earthquake prediction.

[0094] The earthquake precursor prediction effectiveness evaluation and prediction parameter optimization device provided in the embodiment of the present invention has the same implementation principle and technical effects as the aforementioned earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method embodiment. For the sake of brief description, for matters not mentioned in the embodiment of the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization device, reference may be made to the corresponding content in the aforementioned earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method embodiment.

[0095] An embodiment of the present invention provides an electronic device, such as Figure 9As shown, it is a structural diagram of the electronic device, wherein the electronic device includes a processor 91 and a memory 90, the memory 90 stores computer executable instructions that can be executed by the processor 91, and the processor 91 executes the computer executable instructions to implement the above-mentioned earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method.

[0096] exist Figure 9 In the illustrated embodiment, the electronic device further includes a bus 92 and a communication interface 93 , wherein the processor 91 , the communication interface 93 and the memory 90 are connected via the bus 92 .

[0097] Among them, the memory 90 may include high-speed random access memory (RAM), and may also include non-volatile memory (non-volatile memory), such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 93 (which can be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 92 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 92 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0098] Processor 91 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method may be performed by hardware integrated logic circuits within processor 91 or by software instructions. The processor 91 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention may be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules within the decoding processor. The software module may be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. The storage medium is located in the memory, and the processor 91 reads the information in the memory and, in combination with its hardware, completes the steps of the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method of the aforementioned embodiment.

[0099] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0100] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0101] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A method for evaluating earthquake precursor prediction performance and optimizing prediction parameters, characterized in that: include: Preprocessing earthquake precursor signals and earthquake catalog data, and performing grid search on parameter combinations to be optimized to generate multiple different parameter combinations; wherein the parameters to be optimized in the parameter combinations to be optimized include the lower limit of magnitude, epicentral distance, focal depth, lead time, and forecast window length, and different parameter combinations correspond to different numbers of earthquake catalog samples; Predicting the preprocessed data based on a plurality of different parameter combinations using a preset prediction strategy to obtain prediction results corresponding to each parameter combination; wherein the prediction results include the time point of occurrence of an earthquake precursor anomaly event and / or the time point of occurrence of the earthquake event within a preset time period; Based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution, a Molchan diagram corresponding to each parameter combination is generated; wherein the probability distribution represents the probability of successfully predicting different numbers of earthquakes under different alarm rates, and each Molchan diagram includes a corresponding actual prediction curve and a confidence interval. The actual prediction curve represents the first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination, and the confidence intervals corresponding to different Molchan diagrams have the same confidence level; determining a first surface fraction value between the actual prediction curve and the confidence interval in each Molchan plot, and determining an optimal parameter combination from a plurality of different parameter combinations based on the first surface fraction value; Based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution, a Molchan diagram corresponding to each parameter combination is generated, including: generating an actual prediction curve corresponding to each parameter combination based on the prediction results corresponding to each parameter combination; generating a confidence interval corresponding to each parameter combination based on the number of earthquake catalog samples corresponding to each parameter combination and the probability distribution; and forming a corresponding Molchan diagram by combining the actual prediction curve and the confidence interval corresponding to each parameter combination to obtain a Molchan diagram corresponding to each parameter combination; Determining a first area fraction value between an actual prediction curve and a confidence interval in each Molchan diagram includes: determining a third area fraction value between the actual prediction curve and a horizontal axis in each Molchan diagram, and determining a fourth area fraction value between the confidence interval and the horizontal axis in each Molchan diagram; for each Molchan diagram, subtracting the fourth area fraction value corresponding to the Molchan diagram from the third area fraction value corresponding to the Molchan diagram to obtain a first area fraction value corresponding to the Molchan diagram.

2. The earthquake precursor prediction performance evaluation and prediction parameter optimization method according to claim 1, characterized in that: Also includes: Based on the preset second relationship between the alarm rate and the success rate, a reference line is added to the target Molchan diagram corresponding to the optimal parameter combination; Determine a second area fraction value between the actual prediction curve and the reference line in the target Molchan diagram, and evaluate the earthquake precursor prediction effectiveness corresponding to the optimal parameter combination based on the first area fraction value and the second area fraction value corresponding to the target Molchan diagram.

3. The earthquake precursor prediction performance evaluation and prediction parameter optimization method according to claim 1, characterized in that: The pre-processed data is a time series that represents the change of abnormal index values ​​over time within a preset period; The preset prediction strategy includes: for each parameter combination, traversing a preset abnormal threshold set to use each abnormal threshold in the abnormal threshold set to determine whether there is an earthquake precursor abnormal signal in the time series whose abnormal index value exceeds the corresponding abnormal threshold, and determining the time point corresponding to the earthquake precursor abnormal signal as the time point of occurrence of the earthquake precursor abnormal event, then determining a forecast time window within a preset time period based on the determined time point of occurrence of the earthquake precursor abnormal event and the lead time and forecast window length in the parameter combination, and determining whether there is an earthquake event occurrence time point within each forecast time window based on the lower limit of magnitude, epicenter distance and focal depth in the parameter combination; The actual prediction curve uses the alarm rate as an independent variable and the success rate as a dependent variable; based on the prediction results corresponding to each parameter combination, the actual prediction curve corresponding to each parameter combination is generated, including: For each parameter combination, the alarm rate and success rate of the parameter combination corresponding to each abnormal threshold are determined based on the time point of occurrence of the earthquake precursor abnormal event and / or the time point of occurrence of the earthquake event corresponding to the parameter combination, and the actual prediction curve corresponding to the parameter combination is generated based on the alarm rate and success rate corresponding to the parameter combination.

4. The earthquake precursor prediction performance evaluation and prediction parameter optimization method according to claim 1, characterized in that: The probability distribution adopts binomial distribution; based on the number of earthquake catalog samples corresponding to each parameter combination and the probability distribution, the confidence interval corresponding to each parameter combination is generated, including: Based on the number of earthquake catalog samples corresponding to each parameter combination and the binomial distribution, the range of earthquake numbers successfully predicted for each parameter combination at different alarm rates is determined; wherein the sum of the probability sizes of the earthquake number ranges corresponding to each alarm rate matches the confidence level of the corresponding confidence interval; Based on the number of earthquake catalog samples and the range of earthquake numbers corresponding to each parameter combination, the confidence interval corresponding to each parameter combination is determined.

5. The earthquake precursor prediction performance evaluation and prediction parameter optimization method according to claim 1, characterized in that: Perform a grid search on the parameter combinations to be optimized to generate multiple different parameter combinations, including: Based on the search range and search step size of each parameter to be optimized, construct a parameter grid of the parameter combination to be optimized; wherein the parameter combination is represented by grid points in the parameter grid; The parameter grid is traversed to extract the parameter combination represented by each grid point in the parameter grid, thereby obtaining a plurality of different parameter combinations.

6. The earthquake precursor prediction performance evaluation and prediction parameter optimization method according to claim 4, characterized in that: The confidence level is 95%; the calculation formula for the first area fraction value is: in, is the first surface integral value, For the success rate, is the upper limit of the confidence interval, is the alarm rate.

7. An earthquake precursor prediction performance evaluation and prediction parameter optimization device, characterized in that: include: a processing module for preprocessing earthquake precursor signals and earthquake catalog data, and performing a grid search on parameter combinations to be optimized to generate a plurality of different parameter combinations; wherein the parameters to be optimized in the parameter combinations to be optimized include a magnitude lower limit, epicentral distance, focal depth, lead time, and prediction window length, and different parameter combinations correspond to different numbers of earthquake catalog samples; A prediction module is configured to predict the preprocessed data based on a plurality of different parameter combinations using a preset prediction strategy, and obtain prediction results corresponding to each parameter combination; wherein the prediction results include the time point of occurrence of an abnormal earthquake precursor event and / or the time point of occurrence of the earthquake event within a preset time period; A generation module is configured to generate a Molchan diagram corresponding to each parameter combination based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution; wherein the probability distribution represents the probability of successfully predicting different numbers of earthquakes at different alarm rates, and each Molchan diagram includes a corresponding actual prediction curve and a confidence interval, wherein the actual prediction curve represents a first relationship between the alarm rate and the success rate corresponding to the corresponding parameter combination, and the confidence intervals corresponding to different Molchan diagrams have the same confidence level; a determination module, configured to determine a first surface fraction value between an actual prediction curve and a confidence interval in each Molchan diagram, and determine an optimal parameter combination from a plurality of different parameter combinations based on the first surface fraction value; Based on the prediction results corresponding to each parameter combination, the number of earthquake catalog samples, and a preset probability distribution, a Molchan diagram corresponding to each parameter combination is generated, including: generating an actual prediction curve corresponding to each parameter combination based on the prediction results corresponding to each parameter combination; generating a confidence interval corresponding to each parameter combination based on the number of earthquake catalog samples corresponding to each parameter combination and the probability distribution; and forming a corresponding Molchan diagram by combining the actual prediction curve and the confidence interval corresponding to each parameter combination to obtain a Molchan diagram corresponding to each parameter combination; Determining a first area fraction value between an actual prediction curve and a confidence interval in each Molchan diagram includes: determining a third area fraction value between the actual prediction curve and a horizontal axis in each Molchan diagram, and determining a fourth area fraction value between the confidence interval and the horizontal axis in each Molchan diagram; for each Molchan diagram, subtracting the fourth area fraction value corresponding to the Molchan diagram from the third area fraction value corresponding to the Molchan diagram to obtain a first area fraction value corresponding to the Molchan diagram.

8. An electronic device, characterized in that: It includes a processor and a memory, the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the earthquake precursor prediction effectiveness evaluation and prediction parameter optimization method according to any one of claims 1 to 6.

Images