An artificial intelligence-based template matching method for earthquake detection

Abstract

The present invention discloses an artificial intelligence-based template matching earthquake detection method, and the steps include: extracting effective seismic phase data by using a deep learning detection model; using the effective seismic waveform as a machine-detected seismic waveform template; intercepting the artificial seismic monitoring waveform as an artificially-detected seismic waveform template; jointly building a template seismic waveform library; obtaining the template waveforms of each seismic phase; intercepting the continuous waveform data suspected of having an earthquake event as the continuous waveform to be detected, and performing matching calculations between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library. By performing matching calculations between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library, the template matching earthquake detection method can quickly and accurately identify and judge the continuous waveform to be detected, so as to quickly determine whether there is an earthquake event in the continuous waveform to be detected.

Description

Technical Field

[0001] The present invention relates to a seismic detection method, in particular to a template matching seismic detection method based on artificial intelligence. Background Art

[0002] Accurate seismic source parameters are an important prerequisite for studying seismic active structures and buried faults. There are mainly two categories of traditional seismic detection techniques. One is the seismic detection technique based on seismic phase picking. Common methods include manual picking method, long-short window method, and machine learning method, etc. Such methods integrate techniques of arrival time picking, seismic phase association, and earthquake location. They can identify and detect a large number of earthquakes. However, when the waveform has a low signal-to-noise ratio or there are many other seismic phase interferences, the accuracy of seismic phase picking of this type of method will decrease.

[0003] The other is the seismic detection technique based on waveform. Commonly used ones include MFT matching filtering method, M&L matching location method, and GPU-accelerated detection technique based on image. Such methods mainly use waveform cross-correlation technology. They use the waveforms in a specified time window to scan continuous waveforms and determine whether there is a detected earthquake according to the magnitude of the cross-correlation coefficient. The lower magnitude limit of earthquake detection of this type of method is smaller, especially suitable for the detection of repeated earthquakes. However, it has a high generalization ability for template matching and the overall calculation is time-consuming, and it is more suitable for waveform detection with a low signal-to-noise ratio.

[0004] In summary, although the existing two categories of techniques have been widely used in the field of seismic detection and location, neither the seismic detection technique based on seismic phase nor the seismic detection technique based on waveform can single-handedly meet the requirements of fast and accurate seismic detection, and cannot solve the problem of the imbalance between the accuracy of seismic phase picking and the speed of seismic picking in the existing technology. Summary of the Invention

[0005] The object of the invention is to provide a template matching seismic detection method based on artificial intelligence to solve the problem of the imbalance between the accuracy and the detection speed of seismic detection in the existing technology.

[0006] Technical Solution: The template matching seismic detection method based on artificial intelligence of the present invention includes the following steps:

[0007] Step 1, obtain the original continuous waveform data of historical seismic monitoring of each seismograph station, and extract the effective seismic phase data from the original continuous waveform data by using a trained deep learning detection model;

[0008] Step 2, correlate the extracted effective seismic phase data of each to determine each effective seismic waveform, and use each effective seismic waveform as a template for machine-detected seismic waveforms;

[0009] Step 3: Obtain the identified artificial earthquake catalog, intercept the artificial earthquake monitoring waveforms according to the earthquake events of each seismograph station, and then use the intercepted artificial earthquake monitoring waveforms as the artificial detection earthquake waveform templates respectively;

[0010] Step 4: Merge the machine-detected earthquake waveform templates and the artificial-detected earthquake waveform templates to jointly build a template earthquake waveform library;

[0011] Step 5: Perform an intercepting process on each earthquake waveform template in the template earthquake waveform library to obtain the template waveforms of each seismic phase;

[0012] Step 6: Read the continuous waveform data to be detected, intercept the to-be-detected continuous waveforms suspected of having earthquake events from the continuous waveform data, and perform a matching calculation between the to-be-detected continuous waveforms and the template waveforms of each seismic phase of each earthquake waveform template in the template earthquake waveform library. If the calculation result shows that the waveform matching is successful, it is determined that there is an earthquake event in the to-be-detected continuous waveform data; if the calculation result shows that the waveform matching fails, it is determined that there is no earthquake event in the to-be-detected continuous waveform data.

[0013] Further, in Step 1, the specific steps of using the trained deep learning detection model to extract the effective seismic phase data from the original continuous waveform data are as follows:

[0014] Step 1.1: Set the time window length and step size for segmenting and intercepting the original continuous waveform data, and then sequentially intercept the original continuous waveform data of each seismograph station according to the time window length and step size to obtain each segmented waveform of each seismograph station;

[0015] Step 1.2: Input each segmented waveform into the trained deep learning detection model in sequence, and use the deep learning detection model to extract and calculate the probability values of the P seismic phase and the S seismic phase in each segmented waveform;

[0016] Step 1.3: Analyze and judge the probability values of the P seismic phase and the S seismic phase in each segmented waveform: If the probability values of the P seismic phase and the S seismic phase of a certain segmented waveform are both greater than the segmented preset threshold, it is determined that the current segmented waveform is effective seismic phase data; otherwise, it is determined that there is no effective seismic phase data in the current segmented waveform;

[0017] Step 1.4: Repeat Step 1.3 until all segmented waveforms of each seismograph station have completed the analysis and judgment, and respectively output all the effective seismic phase data of each seismograph station, and the effective seismic phase data of each seismograph station are arranged in chronological order.

[0018] Further, in Step 2, the specific steps of correlating the extracted effective seismic phase data to determine each effective earthquake waveform are as follows:

[0019] Step 2.1: Obtain the seismometer station seismic point coordinates corresponding to each valid seismic phase data and the acquisition time of each valid seismic phase data, and then set a seismic source limiting circle for limiting the possible range of the seismic source, an extraction time interval for defining the time range of data extraction, and the length of the valid waveform interception time;

[0020] Step 2.2: Allocate the valid seismic phase data of each seismometer station to the extraction time interval where the corresponding moment is located in sequence according to the acquisition time order, and then form each valid seismic phase data within each extraction time interval into a seismic phase data group;

[0021] Step 2.3: Take out a seismic phase data group in sequence according to the arranged time order, and decompose the P seismic phase wave and S seismic phase wave in each valid seismic phase data in the current seismic phase data group;

[0022] Step 2.4: Determine whether there is still unextracted valid seismic phase data in the seismic phase data group. If there is still unextracted valid seismic phase data, take out a valid seismic phase data and enter Step 2.5; otherwise, enter Step 2.6;

[0023] Step 2.5: Judge the correlation between the P seismic phase wave and S seismic phase wave in the taken-out valid seismic phase data: If the arrival time of the S seismic phase wave in the valid seismic phase data of a certain seismometer station is greater than the arrival time of the P seismic phase wave, then determine that the current valid seismic phase data is possible event data, and then return to Step 2.4; otherwise, determine that the current valid seismic phase data is invalid event data, delete the current valid seismic phase data, and then return to Step 2.4;

[0024] Step 2.6: Determine whether the number of possible event data in the current seismic phase data group is greater than the event number threshold. If the number is less than the event number threshold, then determine that the current seismic phase data group is an invalid data group, and then enter Step 2.10; otherwise, enter Step 2.7;

[0025] Step 2.7: Calculate the distance D between each seismometer station and the seismic source according to the arrival time of the S seismic phase wave, the arrival time of the P seismic phase wave, the S seismic phase wave velocity, and the P seismic phase wave velocity in each possible event data respectively, and then draw each seismic source positioning circle with the seismometer station seismic point coordinates of each seismometer station as the center and the corresponding distance D as the radius. After drawing the seismic source positioning circles corresponding to each possible event data, mark the intersection points of each seismic source positioning circle as the positioning reference points;

[0026] Step 2.8: Select each positioning reference point as the center of the seismic source limiting circle in sequence, then determine the number of points of the positioning reference points within each seismic source limiting circle respectively, and then compare the determined number of points of each to find the maximum number of points. If the maximum number of points is greater than the point number threshold, then determine that the current seismic phase data group is a valid data group, and then enter Step 2.9; otherwise, determine that the current seismic phase data group is an invalid data group, and then enter Step 2.10;

[0027] Step 2.9: Find the source-limiting circle corresponding to the maximum number of points, then determine each positioning reference point within the source-limiting circle, then find the corresponding source-locating circle based on each positioning reference point, then determine each seismograph station according to each source-locating circle, and finally intercept the effective seismic waveform from the effective phase data of each seismograph station according to the time length of the effective waveform interception. The effective seismic waveform includes the complete waveforms of S-phase waves and P-phase waves.

[0028] Step 2.10: Determine whether there are still unextracted phase data groups. If there are still unextracted phase data groups, return to Step 2.3; otherwise, output the effective seismic waveforms of each seismograph station.

[0029] Furthermore, in Step 3, the specific steps for intercepting artificial earthquake monitoring waveforms according to the earthquake events of each seismograph station are as follows:

[0030] Step 3.1: Determine the earthquake events of each seismograph station according to the identified artificial earthquake catalog, and obtain the artificial monitoring data of each earthquake event.

[0031] Step 3.2: Use the TAUP ray tracing technique to calculate the arrival times of S-phase waves and P-phase waves from each earthquake event to each seismograph station.

[0032] Step 3.3: Set the starting point of the effective waveform interception interval to 7 seconds before the arrival time of the P-phase wave, then set the ending point of the effective waveform interception interval to 15 seconds after the arrival time of the P-phase wave, and then intercept the corresponding artificial earthquake monitoring waveforms from the artificial monitoring data of each seismograph station according to the set effective waveform interception interval.

[0033] Furthermore, in Step 4, the specific steps for merging the machine-detected seismic waveform template and the manually-detected seismic waveform template to jointly build a template seismic waveform library are as follows:

[0034] Step 4.1: Collect each machine-detected seismic waveform template and each manually-detected seismic waveform template, and obtain the seismograph stations corresponding to the machine-detected seismic waveform templates and the seismograph stations corresponding to the manually-detected seismic waveform templates.

[0035] Step 4.2: Use the names of each seismograph station as the first-level directories of the template seismic waveform library, and place each machine-detected seismic waveform template and each manually-detected seismic waveform template under the corresponding first-level directory.

[0036] Step 4.3: Extract the P-phase waves from the machine-detected seismic waveform templates or manually-detected seismic waveform templates under each first-level directory. Classify the obtained seismic waveform templates according to the amplitude of the P-phase waves, and use the classification level as the second-level directory. Then, place each machine-detected seismic waveform template and each manually-detected seismic waveform template into the corresponding second-level directory. If there are multiple seismic waveform templates under a certain level, delete the redundant seismic waveform templates so that only one seismic waveform template is retained for each level, ultimately constituting the template seismic waveform library.

[0037] Further, in step 4.3, the specific steps for classifying the obtained seismic waveform templates according to the amplitude of the P-phase waves are as follows:

[0038] First, decompose each machine-detected seismic waveform template and each manually-detected seismic waveform template collected in step 4.1 to obtain the P-phase waves in each seismic waveform template;

[0039] Then, screen out the maximum amplitude value among each P-phase wave and define it as Tmax, and screen out the minimum amplitude value among each P-phase wave and define it as Tmin;

[0040] Then, set 0.8Tmin as the minimum level value, set 1.6Tmax as the maximum level value, and equally-spacedly set 8 level values between 0.8Tmin and 1.6Tmax. Use the minimum level value, the maximum level value, and the 8 level values in between to divide the amplitude of the P-phase waves into 10 level intervals;

[0041] Finally, determine which level among the 10 level intervals the amplitude of the P-phase wave belongs to according to the amplitude of the P-phase wave, which is the classification level of the corresponding seismic waveform template.

[0042] Further, in step 5, the specific steps for intercepting each seismic waveform template in the template seismic waveform library to obtain the template waveforms of each seismic phase are as follows:

[0043] Step 5.1: Obtain a seismic waveform template under a second-level directory in the template seismic waveform library, and separate the P-phase wave and the S-phase wave from the obtained seismic waveform template;

[0044] Step 5.2: Respectively intercept the separated P-phase wave and S-phase wave: Use the waveform within the interval from 1 s before to 5 s after the P-phase wave as the P-phase wave template waveform, and use the waveform within the interval from 1 s before to 5 s after the S-phase wave as the S-phase wave template waveform;

[0045] Step 5.3: Replace the seismic waveform templates in the corresponding subdirectories with the obtained P-phase wave template waveforms and S-phase wave template waveforms, and then determine whether there are still seismic waveform templates in the template seismic waveform library that have not been intercepted. If there are still seismic waveform templates that have not been intercepted, return to Step 5.1; otherwise, proceed to Step 5.4.

[0046] Step 5.4: Update the template seismic waveform library: Name the P-phase wave template waveforms in each subdirectory as P waves, and name the S-phase wave template waveforms in each subdirectory as S waves.

[0047] Furthermore, in Step 6, the specific steps for intercepting the continuous waveform data suspected of containing seismic events from the continuous waveform data are as follows:

[0048] Step 6.1.1: Extract each data segment with waveforms in the continuous waveform data, and then perform waveform analysis on each intercepted data segment to obtain the suspected waveforms of each data segment.

[0049] Step 6.1.2: Use the deep learning detection model to identify the separated suspected waveforms, and determine whether each suspected waveform is a P-phase waveform or an S-phase waveform. If P-phase waveforms and S-phase waveforms are identified, proceed to Step 6.1.3; otherwise, determine that there are no seismic events in the continuous waveform data, and then proceed to Step 6.1.6.

[0050] Step 6.1.3: Conduct statistical judgment on the identified P-phase waveforms and S-phase waveforms. If there is one P-phase waveform and one S-phase waveform, proceed to Step 6.1.4; if there are multiple P-phase waveforms and S-phase waveforms, proceed to Step 6.1.5.

[0051] Step 6.1.4: Match the waveforms of the S-phase wave and the P-phase wave. If the S-phase wave and the P-phase wave belong to the same seismic source and the arrival time of the S-phase wave is greater than the arrival time of the P-phase wave, determine that there is a suspected seismic event in the continuous waveform data. Then, splice the S-phase wave and the P-phase wave in chronological order according to the arrival time of the S-phase wave and the P-phase wave to form the continuous waveform to be inspected, and then proceed to Step 6.1.6; otherwise, determine that there are no seismic events in the continuous waveform data, and then proceed to Step 6.1.6.

[0052] Step 6.1.5: Perform waveform matching on each P-phase waveform and each S-phase waveform. The matched P-phase waveform and S-phase waveform are regarded as a seismic phase matching group. If the number of obtained seismic phase matching groups ≥ 1, it is determined that there may be a seismic event in the continuous waveform data. Then, the S-phase waves and P-phase waves in each seismic phase matching group are respectively spliced into each continuous waveform to be detected in the time order of the arrival time of the S-phase wave and the arrival time of the P-phase wave, and then go to Step 6.1.6. Otherwise, it is determined that there is no seismic event in the continuous waveform data, and then go to Step 6.1.6;

[0053] Step 6.1.6: Output the determination result. If it is determined that there may be a seismic event in the continuous waveform data, output each continuous waveform to be detected obtained by splicing. If it is determined that there is no seismic event in the continuous waveform data, output the result of not obtaining the continuous waveform to be detected.

[0054] Further, in Step 6.1.5, the specific steps for performing waveform matching on each P-phase waveform and each S-phase waveform are as follows:

[0055] Step a: Perform filtering processing on each P-phase waveform and each S-phase waveform to obtain the denoised P-phase waveform and S-phase waveform;

[0056] Step b: Arrange each S-phase waveform in ascending order of arrival time to form an S-wave sequence, and arrange each P-phase waveform in ascending order of arrival time to form a P-wave sequence;

[0057] Step c: Take out the S-phase waveform with the smallest arrival time from the S-wave sequence, and then find out each P-phase waveform in the P-wave sequence whose arrival time is smaller than the taken-out S-phase waveform. If no corresponding P-phase waveform is found, go to Step h;

[0058] Step d: Obtain the maximum amplitude of the taken-out S-phase waveform as Hs and the maximum amplitudes of each of the found P-phase waveforms as Hp. Then, set the height of the intercept line to 20%Hs. Use the intercept line with a height of 20%Hs to perform waveform interception on the taken-out S-phase waveform and the found P-phase waveforms, obtain the distribution time width ΔS of the S-phase waveform above the intercept line and the distribution time width ΔP of each P-phase waveform, and then count the number of times Ns that the S-phase waveform crosses the intercept line up and down within the range of the distribution time width ΔS and the number of times Np that each corresponding P-phase waveform crosses the intercept line up and down within the range of each distribution time width ΔP;

[0059] Step e: Find out the P-phase waveforms with Hp < Hs among each P-phase waveform, and regard the found P-phase waveforms as the P-phase waveforms after the first-level screening. If no corresponding P-phase waveform is found, go to Step h;

[0060] Step f: Among the P-phase waveforms after the first-level screening, find each P-phase waveform with 0.85 < Np / Ns < 1.15, and use the found P-phase waveforms as the P-phase waveforms after the second-level screening. If no corresponding P-phase waveforms are found, go to Step h;

[0061] Step g: Among the P-phase waveforms after the second-level screening, find the distribution time width ΔP with the smallest difference from the distribution time width ΔS, and determine the P-phase waveform corresponding to the found distribution time width ΔP as the one that matches the current S-phase waveform. Then delete the found P-phase waveform from the P-wave sequence, and go to Step h;

[0062] Step h: Determine whether there are still S-phase waveforms in the S-wave sequence. If there are still S-phase waveforms, return to Step c; otherwise, go to Step i;

[0063] Step i: Output the matched P-phase waveforms and S-phase waveforms in pairs.

[0064] Further, in Step 6, the specific steps for matching and calculating the continuous waveform to be detected with the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library are as follows:

[0065] Step 6.2.1: Determine the seismograph station to which the continuous waveform to be detected belongs based on the data source of the continuous waveform to be detected, and then determine the first-level directory in the template seismic waveform library according to the seismograph station;

[0066] Step 6.2.2: Calculate the maximum amplitude of the P-phase waveform in the continuous waveform to be detected, and then match the corresponding grade classification according to the maximum amplitude of the P-phase waveform, so as to determine the second-level directory under the first-level directory. Then determine whether there are P-waves and S-waves in the determined second-level directory. If there are P-waves and S-waves, go to Step 6.2.3; otherwise, go to Step 6.2.4;

[0067] Step 6.2.3: Determine the maximum amplitude S1 of the S-phase waveform in the continuous waveform to be detected and the maximum amplitude S2 of the S-wave in the second-level directory, and then calculate the ratio of the maximum amplitude S1 to the maximum amplitude S2. If the ratio is within the range of 0.9 to 1.1, it is determined that the current continuous waveform to be detected is successfully matched; otherwise, it is determined that the current continuous waveform to be detected is failed to be matched, and then go to Step 6.2.6;

[0068] Step 6.2.4: Search for P-waves and S-waves in the second-level directories of adjacent levels step by step, and enter Step 6.2.5 after finding P-waves and S-waves in the second-level directories of adjacent levels;

[0069] Step 6.2.5: Determine the maximum amplitude P1 of the P-phase waveform and the maximum amplitude S1 of the S-phase waveform in the continuous waveform to be detected. Then, determine the maximum amplitude P2 of the P-wave and the maximum amplitude S2 of the S-wave in the second-level directory of the adjacent level. Next, calculate the matching degree G as P1×S2 / P2×S1, and then determine whether the matching degree G is within the matching threshold range of 0.8 to 1.2. If it is within the matching threshold range, it is determined that the current continuous waveform to be detected is successfully matched, and the P-phase waveform and the S-phase waveform in the continuous waveform to be detected are respectively stored as the P-wave and the S-wave in the second-level directory that matches the P-phase waveform level in the template seismic waveform library. Otherwise, it is determined that the current continuous waveform to be detected fails to match, and then proceed to Step 6.2.6;

[0070] Step 6.2.6: Output the result of the matching calculation.

[0071] Compared with the prior art, the beneficial effects of the present invention are as follows: By combining the machine-detected seismic waveform template and the manually-detected seismic waveform template to jointly build the template seismic waveform library, the template seismic waveform library can have a more comprehensive database, effectively improving the success rate and reliability of the seismic detection waveform template matching; By performing matching calculations on the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library with the continuous waveform to be detected, the recognition and judgment of the continuous waveform to be detected can be quickly and accurately realized, so as to quickly determine whether there is a seismic event in the continuous waveform to be detected; By intercepting each seismic waveform template in the template seismic waveform library, the amount of data for comparison can be reduced, the comparison of invalid data can be reduced, making the template matching more accurate and reliable, and also improving the template comparison efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is the flowchart of the method of the present invention;

[0073] Figure 2 is the schematic diagram of the extraction time interval of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0074] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the described embodiments.

[0075] As Figure 1 shown, the template matching seismic detection method based on artificial intelligence disclosed by the present invention includes the following steps:

[0076] Step 1: Obtain the original continuous waveform data of historical seismic monitoring of each seismograph station, and use the trained deep learning detection model to extract the effective phase data from the original continuous waveform data;

[0077] Step 2: Correlate the extracted effective seismic phase data to determine each effective seismic waveform, and use each effective seismic waveform as a template for machine-detected seismic waveforms.

[0078] Step 3: Obtain the recognized artificial earthquake catalog, intercept the artificial earthquake monitoring waveforms according to the earthquake events of each seismograph station, and use the intercepted artificial earthquake monitoring waveforms as templates for each artificial-detected seismic waveform.

[0079] Step 4: Merge the template for machine-detected seismic waveforms and the template for artificial-detected seismic waveforms to jointly build a template seismic waveform library.

[0080] Step 5: Perform an intercepting process on each seismic waveform template in the template seismic waveform library to obtain the template waveforms of each seismic phase.

[0081] Step 6: Read the continuous waveform data to be detected, intercept the continuous waveform to be detected suspected of having an earthquake event from the continuous waveform data, and perform a matching calculation between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library. If the calculation result shows that the waveform matching is successful, it is determined that there is an earthquake event in the continuous waveform data to be detected; if the calculation result shows that the waveform matching fails, it is determined that there is no earthquake event in the continuous waveform data to be detected.

[0082] Using the template for machine-detected seismic waveforms and the template for artificial-detected seismic waveforms to jointly build a template seismic waveform library can make the template seismic waveform library have a more comprehensive database, effectively improving the success rate and reliability of seismic detection waveform template matching; by performing a matching calculation between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library, the recognition and judgment of the continuous waveform to be detected can be quickly and accurately realized, so as to quickly determine whether there is an earthquake event in the continuous waveform to be detected; by performing an intercepting process on each seismic waveform template in the template seismic waveform library, the amount of data for comparison can be reduced, the comparison of invalid data can be reduced, making the template matching more accurate and reliable, and also improving the template comparison efficiency.

[0083] Further, in Step 1, the specific steps for using the trained deep learning detection model to extract the effective seismic phase data from the original continuous waveform data are as follows:

[0084] Step 1.1: Set the time window length and step length for segmenting and intercepting the original continuous waveform data, and then sequentially intercept the original continuous waveform data of each seismograph station according to the time window length and step length to obtain each segmented waveform of each seismograph station.

[0085] Step 1.2: Input each segmented waveform into the trained deep learning detection model in sequence. Use the deep learning detection model to extract and calculate the probability values of the P-phase and S-phase in each segmented waveform. The deep learning detection model technology is an existing conventional technology. Through the deep learning detection model, the probability value indicating whether the segmented waveform belongs to the P-phase or S-phase can be quickly output, thus realizing the fast and intelligent classification of the segmented waveforms.

[0086] Step 1.3: Analyze and judge the probability values of the P-phase and S-phase in each segmented waveform. If the probability values of the P-phase and S-phase of a certain segmented waveform are both greater than the segmented preset threshold, it is determined that the current segmented waveform is valid seismic phase data; otherwise, it is judged that there is no valid seismic phase data in the current segmented waveform.

[0087] Step 1.4: Repeat Step 1.3 until all segmented waveforms of each seismograph station have completed the analysis and judgment, and respectively output all the valid seismic phase data of each seismograph station, and the valid seismic phase data of each seismograph station are arranged in chronological order.

[0088] Furthermore, the deep learning detection model is a PhaseNet convolutional neural network deep learning model. When training the deep learning detection model:

[0089] First, obtain the existing earthquake data training set and earthquake data test set. Directly using the existing earthquake data to construct the training set and test set can make the trained model have better reliability.

[0090] Then, use the earthquake data training set to train the deep learning detection model.

[0091] Then, use the earthquake data test set to test the trained deep learning detection model, and finally complete the training of the deep learning detection model.

[0092] Furthermore, in Step 2, the specific steps to correlate the extracted valid seismic phase data to determine the specific valid earthquake waveforms are as follows:

[0093] Step 2.1: Obtain the seismograph station measurement point coordinates corresponding to each valid seismic phase data and the acquisition time of each valid seismic phase data. Then set a source limiting circle for limiting the possible range of the earthquake source, an extraction time interval for delimiting the data extraction time range, and the effective waveform interception time length. Preferably, the extraction time interval is 180 s, and the effective waveform interception length is 120 s. By setting the corresponding extraction time interval, the relevant valid seismic phase data can be screened once in terms of time.

[0094] Step 2.2: Allocate the effective seismic phase data of each seismograph station to the corresponding extraction time intervals where the corresponding moments are arranged in chronological order of acquisition time, and then form the effective seismic phase data of each seismograph station within each extraction time interval into a seismic phase data group.

[0095] Step 2.3: Sequentially extract a seismic phase data group in the arranged time order, and decompose the P seismic phase wave and S seismic phase wave in each effective seismic phase data in the current seismic phase data group.

[0096] Step 2.4: Determine whether there is still unextracted effective seismic phase data in the seismic phase data group. If there is still unextracted effective seismic phase data, extract an effective seismic phase data and go to Step 2.5; otherwise, go to Step 2.6.

[0097] Step 2.5: Judge the correlation between the P seismic phase wave and S seismic phase wave in the extracted effective seismic phase data: If the arrival time of the S seismic phase wave in the effective seismic phase data of a certain seismograph station is greater than the arrival time of the P seismic phase wave, then determine that the current effective seismic phase data is possible event data, and then return to Step 2.4; otherwise, determine that the current effective seismic phase data is invalid event data, delete the current effective seismic phase data, and then return to Step 2.4.

[0098] Step 2.6: Judge whether the number of possible event data in the current seismic phase data group is greater than the event number threshold. The event number threshold is set to 3. The purpose of setting the event number threshold is on the one hand to avoid accidental errors, and on the other hand to ensure the basic requirements for subsequent earthquake source location. If the number is less than the event number threshold, then determine that the current seismic phase data group is an invalid data group, and then go to Step 2.10; otherwise, go to Step 2.7.

[0099] Step 2.7: Calculate the distance D between each seismograph station and the earthquake source according to the arrival time of the S seismic phase wave, the arrival time of the P seismic phase wave, the S seismic phase wave velocity, and the P seismic phase wave velocity in each possible event data respectively. Then, draw each earthquake source location circle with the seismograph point coordinates of each seismograph station as the center and the corresponding distance D as the radius. After drawing the earthquake source location circles corresponding to each possible event data, mark the intersection points of each earthquake source location circle as the location reference points.

[0100] Step 2.8: Sequentially select each location reference point as the center of the earthquake source limiting circle, then determine the number of points of the location reference points within each earthquake source limiting circle respectively, and then compare the determined number of points to find the maximum number of points. If the maximum number of points is greater than the point number threshold, and the point number threshold is set to 0.8 times the number of possible event data in the current seismic phase data group to ensure the reliability of earthquake source location, then determine that the current seismic phase data group is an effective data group, and then go to Step 2.9; otherwise, determine that the current seismic phase data group is an invalid data group, and then go to Step 2.10.

[0101] Step 2.9: Find the source-limiting circle corresponding to the maximum number of points, then determine each positioning reference point within the source-limiting circle, then find the corresponding source-locating circle based on each positioning reference point, then determine each seismograph station according to each source-locating circle, and finally intercept the effective seismic waveform from the effective phase data of each seismograph station according to the time length of the effective waveform interception. The effective seismic waveform includes the complete waveforms of the S-phase wave and the P-phase wave;

[0102] Step 2.10: Determine whether there are still unextracted phase data groups. If there are still unextracted phase data groups, then return to Step 2.3; otherwise, output the effective seismic waveforms of each seismograph station.

[0103] Furthermore, in Step 3, the specific steps for intercepting the artificial earthquake monitoring waveform according to the earthquake events of each seismograph station are as follows:

[0104] Step 3.1: Determine the earthquake events of each seismograph station according to the identified artificial earthquake catalog, and obtain the artificial monitoring data of each earthquake event;

[0105] Step 3.2: Use the TAUP ray tracing technique to calculate the arrival times of the S-phase wave and the P-phase wave of each earthquake event to each seismograph station respectively;

[0106] Step 3.3: Set the start point of the effective waveform interception interval to 7 seconds before the arrival time of the P-phase wave, then set the end point of the effective waveform interception interval to 15 seconds after the arrival time of the P-phase wave, and then intercept the corresponding artificial earthquake monitoring waveform from the artificial monitoring data of each seismograph station according to the set effective waveform interception interval. By setting the effective waveform interception interval, the data range can be limited and the effective waveform can be retained.

[0107] Furthermore, in Step 4, the specific steps for merging the machine-detected seismic waveform template and the manually-detected seismic waveform template to jointly build the template seismic waveform library are as follows:

[0108] Step 4.1: Collect each machine-detected seismic waveform template and each manually-detected seismic waveform template, and obtain the seismograph station corresponding to the machine-detected seismic waveform template and the seismograph station corresponding to the manually-detected seismic waveform template;

[0109] Step 4.2: Use the names of each seismograph station as the first-level directories of the template seismic waveform library, and place each machine-detected seismic waveform template and each manually-detected seismic waveform template under the corresponding first-level directory;

[0110] Step 4.3: Extract the P-phase waves from the machine-detected seismic waveform templates or the manually-detected seismic waveform templates under each first-level directory. Classify the obtained seismic waveform templates according to the amplitude of the P-phase waves, and use the classified levels as the second-level directories. By setting the level classification, the waveform template matching can be made more accurate. Matching two waveforms with similar amplitudes not only has high matching efficiency but also high matching accuracy. Then, place each machine-detected seismic waveform template and each manually-detected seismic waveform template into the corresponding second-level directory. If there are multiple seismic waveform templates under a certain level, delete the redundant seismic waveform templates so that only one seismic waveform template is retained for each level, finally forming a template seismic waveform library.

[0111] Further, in Step 4.3, the specific steps for classifying the obtained seismic waveform templates according to the amplitude of the P-phase waves are as follows:

[0112] First, decompose each machine-detected seismic waveform template and each manually-detected seismic waveform template collected in Step 4.1 to obtain the P-phase waves in each seismic waveform template;

[0113] Then, screen out the maximum amplitude value among each P-phase wave and define it as Tmax, and screen out the minimum amplitude value among each P-phase wave and define it as Tmin;

[0114] Then set 0.8Tmin as the minimum level value, then set 1.6Tmax as the maximum level value, and then equally spacedly set 8 level values between 0.8Tmin and 1.6Tmax. Divide the amplitude of the P seismic phase wave into 10 level intervals by using the minimum level value, the maximum level value, and the 8 level values in between. Set the amplitude value of the P seismic phase wave as T. Then when 0 < T ≤ 0.8Tmin, determine that the amplitude value is the first level; when 0.8Tmin < T ≤ 0.8Tmin + (1.6Tmax - 0.8Tmin) / 8, determine that the amplitude value is the second level; when 0.8Tmin + (1.6Tmax - 0.8Tmin) / 8 < T ≤ 0.8Tmin + (1.6Tmax - 0.8Tmin) / 4, determine that the amplitude value is the third level; when 0.8Tmin + (1.6Tmax - 0.8Tmin) / 4 < T ≤ 0.8Tmin + 3(1.6Tmax - 0.8Tmin) / 8, determine that the amplitude value is the fourth level; when 0.8Tmin + 3(1.6Tmax - 0.8Tmin) / 8 < T ≤ 0.8Tmin + (1.6Tmax - 0.8Tmin) / 2, determine that the amplitude value is the fifth level; when 0.8Tmin + (1.6Tmax - 0.8Tmin) / 2 < T ≤ 0.8Tmin + 5(1.6Tmax - 0.8Tmin) / 8, determine that the amplitude value is the sixth level; when 0.8Tmin + 5(1.6Tmax - 0.8Tmin) / 8 < T ≤ 0.8Tmin + 3(1.6Tmax - 0.8Tmin) / 4, determine that the amplitude value is the seventh level; when 0.8Tmin + 3(1.6Tmax - 0.8Tmin) / 4 < T ≤ 0.8Tmin + 7(1.6Tmax - 0.8Tmin) / 8, determine that the amplitude value is the eighth level; when 0.8Tmin + 7(1.6Tmax - 0.8Tmin) / 8 < T ≤ 1.6Tmax, determine that the amplitude value is the ninth level; when T > 1.6Tmax, determine that the amplitude value is the tenth level;

[0115] Finally, determine which level among the 10 level intervals the amplitude of the P seismic phase wave belongs to according to the amplitude size of the P seismic phase wave, which is the classification level of the corresponding seismic waveform template.

[0116] Further, in step 5, the specific steps for intercepting each seismic waveform template in the template seismic waveform library to obtain the template waveform of each seismic phase are as follows:

[0117] Step 5.1, obtain the seismic waveform template in a secondary directory of the template seismic waveform library, and separate the P seismic phase wave and the S seismic phase wave from the obtained seismic waveform template;

[0118] Step 5.2, separately intercept the separated P-phase wave and S-phase wave: Take the waveform within the range from the first 1 s to the last 5 s of the P-phase wave as the P-phase wave template waveform, and take the waveform within the range from the first 1 s to the last 5 s of the S-phase wave as the S-phase wave template waveform;

[0119] Step 5.3, use the obtained P-phase wave template waveform and S-phase wave template waveform to replace the seismic waveform template under the corresponding secondary directory, thereby optimizing the original seismic waveform template, and then determine whether there is still a seismic waveform template in the template seismic waveform library that has not been intercepted. If there is still a seismic waveform template that has not been intercepted, return to Step 5.1; otherwise, enter Step 5.4;

[0120] Step 5.4, update the template seismic waveform library: Name the P-phase wave template waveforms under each secondary directory as P waves, and name the S-phase wave template waveforms under each secondary directory as S waves.

[0121] Further, in Step 6, the specific steps for intercepting the continuous waveform to be inspected suspected of containing seismic events from the continuous waveform data are as follows:

[0122] Step 6.1.1, intercept each data segment with waveform in the continuous waveform data, and then perform waveform analysis on each intercepted data segment to obtain the suspected waveforms of each data segment;

[0123] Step 6.1.2, use the deep learning detection model to identify the separated suspected waveforms, and determine whether each suspected waveform is a P-phase waveform or an S-phase waveform. If P-phase waveforms and S-phase waveforms are identified, enter Step 6.1.3; otherwise, determine that there is no seismic event in the continuous waveform data, and then enter Step 6.1.6;

[0124] Step 6.1.3, perform statistical judgment on the identified P-phase waveforms and S-phase waveforms. If there is one P-phase waveform and one S-phase waveform, enter Step 6.1.4; if there are multiple P-phase waveforms and S-phase waveforms, enter Step 6.1.5;

[0125] Step 6.1.4, perform waveform matching on the S-phase wave and the P-phase wave. If the S-phase wave and the P-phase wave belong to the same seismic source and the arrival time of the S-phase wave is greater than that of the P-phase wave, determine that there is a suspected seismic event in the continuous waveform data, and then splice the S-phase wave and the P-phase wave in chronological order according to the arrival time of the S-phase wave and the P-phase wave into the continuous waveform to be inspected, and then enter Step 6.1.6; otherwise, determine that there is no seismic event in the continuous waveform data, and then enter Step 6.1.6;

[0126] Step 6.1.5: Perform waveform matching on each P-phase waveform and each S-phase waveform, and regard the matched P-phase waveform and S-phase waveform as a seismic phase matching group. If the number of obtained seismic phase matching groups ≥ 1, it is determined that there may be a seismic event in the continuous waveform data. Then, splice the S-phase waves and P-phase waves in each seismic phase matching group according to the time sequence of the arrival time of the S-phase wave and the arrival time of the P-phase wave to obtain each continuous waveform to be detected, and then proceed to Step 6.1.6. Otherwise, it is determined that there is no seismic event in the continuous waveform data, and then proceed to Step 6.1.6;

[0127] Step 6.1.6: Output the determination result. If it is determined that there may be a seismic event in the continuous waveform data, output each continuous waveform to be detected obtained by splicing. If it is determined that there is no seismic event in the continuous waveform data, output the result of not obtaining the continuous waveform to be detected.

[0128] Furthermore, in Step 6.1.5, the specific steps for performing waveform matching on each P-phase waveform and each S-phase waveform are as follows:

[0129] Step a: Perform filtering processing on each P-phase waveform and each S-phase waveform to obtain the denoised P-phase waveform and S-phase waveform;

[0130] Step b: Arrange each S-phase waveform in ascending order of arrival time to form an S-wave sequence, and arrange each P-phase waveform in ascending order of arrival time to form a P-wave sequence;

[0131] Step c: Take out the S-phase waveform with the smallest arrival time from the S-wave sequence, and then find each P-phase waveform in the P-wave sequence whose arrival time is smaller than the taken-out S-phase waveform. If no corresponding P-phase waveform is found, proceed to Step h;

[0132] Step d: Obtain the maximum amplitude Hs of the extracted S-phase waveform and the maximum amplitudes Hp of each of the found P-phase waveforms. Then, set the height of the intercept line to 20%Hs. Use the intercept line with a height of 20%Hs to intercept the extracted S-phase waveform and each of the found P-phase waveforms, and obtain the distribution time width ΔS of the S-phase waveform above the intercept line and the distribution time width ΔP of each P-phase waveform. Using the intercept line for waveform interception can determine the effective duration of the waveform. The feature used is that the vibration durations of the S-phase waveform and the P-phase waveform from the same seismic source are close. Then, count the number of times Ns that the S-phase waveform crosses the intercept line up and down within the range of the distribution time width ΔS and the number of times Np that each corresponding P-phase waveform crosses the intercept line up and down within the range of each distribution time width ΔP. Use the number of times of crossing the intercept line combined with the distribution time width to determine whether the vibration frequencies of the S-phase waveform and the P-phase waveform are close, based on the feature that the vibration frequencies of the S-phase waveform and the P-phase waveform from the same seismic source are close;

[0133] Step e: Find the P-phase waveforms in which Hp < Hs among each P-phase waveform, and use the found P-phase waveforms as the P-phase waveforms after the first-level screening. If no corresponding P-phase waveforms are found, go to step h;

[0134] Step f: Find each P-phase waveform in which 0.85 < Np / Ns < 1.15 among the P-phase waveforms after the first-level screening, and use the found P-phase waveforms as the P-phase waveforms after the second-level screening. If no corresponding P-phase waveforms are found, go to step h;

[0135] Step g: Find the distribution time width ΔP with the smallest difference from the distribution time width ΔS among the P-phase waveforms after the second-level screening, and determine the P-phase waveform corresponding to the found distribution time width ΔP as the one that matches the current S-phase waveform. Then, delete the found P-phase waveform from the P-wave sequence, and go to step h;

[0136] Step h: Determine whether there are still S-phase waveforms in the S-wave sequence. If there are still S-phase waveforms, return to step c; otherwise, go to step i;

[0137] Step i: Output the matching P-phase waveforms and S-phase waveforms in pairs.

[0138] Further, in step 6, the specific steps for matching and calculating the continuous waveform to be detected with the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library are as follows:

[0139] Step 6.2.1: Determine the seismograph station to which the continuous waveform to be detected belongs according to the data source of the continuous waveform to be detected, and then determine the first-level directory in the template seismic waveform library according to the seismograph station;

[0140] Step 6.2.2, calculate the maximum amplitude of the P phase waveform in the continuous waveform to be detected, and then match the corresponding grade classification according to the maximum amplitude of the P phase waveform, so as to determine the secondary directory under the primary directory, and then determine whether there are P waves and S waves in the determined secondary directory. If there are P waves and S waves, go to step 6.2.3, otherwise go to step 6.2.4;

[0141] Step 6.2.3, determine the maximum amplitude S1 of the S seismic phase waveform in the continuous waveform to be tested and the maximum amplitude S2 of the S wave in the secondary directory, and then calculate the ratio of the maximum amplitude S1 to the maximum amplitude S2. If the ratio is within the range of 0.9 to 1.1, it is determined that the current continuous waveform to be tested is matched successfully, otherwise it is determined that the current continuous waveform to be tested fails to match, and then proceed to step 6.2.6;

[0142] Step 6.2.4, search for P wave and S wave in the secondary directory of the adjacent level step by step, and enter step 6.2.5 after finding P wave and S wave in the secondary directory of the adjacent level. By searching for P wave and S wave in the secondary directory of the adjacent level, it can be ensured that each input continuous waveform to be tested can complete the template matching;

[0143] Step 6.2.5, determine the maximum amplitude P1 of the P seismic phase waveform and the maximum amplitude S1 of the S seismic phase waveform in the continuous waveform to be tested, then determine the maximum amplitude P2 of the P wave and the maximum amplitude S2 of the S wave in the secondary directory of the adjacent level, then calculate the matching degree G as P1×S2 / P2×S1, and then judge whether the matching degree G is within the matching threshold range of 0.8~1.2. By setting a relatively large range, the errors found due to different levels can be enhanced, and the inclusiveness of template matching can be enhanced. If it is within the matching threshold range, it is determined that the current continuous waveform to be tested is matched successfully, and the P seismic phase waveform and the S seismic phase waveform in the continuous waveform to be tested are stored as P waves and S waves in the secondary directory matching the P seismic phase waveform level in the template seismic waveform library, otherwise it is determined that the current continuous waveform to be tested fails to match, and then enter step 6.2.6;

[0144] Step 6.2.6, output the matching calculation results.

[0145] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. An artificial intelligence-based template matching seismic detection method, characterized in that, Including the following steps: Step 1: Obtain the original continuous waveform data of historical earthquake monitoring at each seismograph station, and use the trained deep learning detection model to extract the effective seismic phase data from the original continuous waveform data; Step 2: Correlate the extracted effective seismic phase data of each station to determine the effective seismic waveforms among them, and use each effective seismic waveform as a template for machine-detected seismic waveforms; Step 3: Obtain the identified artificial earthquake catalog, intercept the artificial earthquake monitoring waveforms according to the earthquake events at each seismograph station, and use each intercepted artificial earthquake monitoring waveform as a template for each artificially detected seismic waveform; Step 4: Merge the template of machine-detected seismic waveforms and the template of artificially detected seismic waveforms to jointly build a template seismic waveform library; Step 5: Perform an intercepting process on each seismic waveform template in the template seismic waveform library to obtain the template waveforms of each seismic phase; Step 6: Read the continuous waveform data to be detected, intercept the continuous waveform to be detected suspected of having an earthquake event from the continuous waveform data, and perform a matching calculation between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library. If the calculation result shows that the waveform matching is successful, it is determined that there is an earthquake event in the continuous waveform data to be detected. If the calculation result shows that the waveform matching fails, it is determined that there is no earthquake event in the continuous waveform data to be detected.

2. The template matching earthquake detection method based on artificial intelligence according to claim 1, wherein In Step 1, the specific steps of using the trained deep learning detection model to extract the effective seismic phase data from the original continuous waveform data are as follows: Step 1.1: Set the time window length and step size for segmenting and intercepting the original continuous waveform data, and then sequentially intercept the original continuous waveform data of each seismograph station according to the time window length and step size to obtain each segmented waveform of each seismograph station; Step 1.2: Sequentially input each segmented waveform into the trained deep learning detection model, and use the deep learning detection model to extract and calculate the probability values of the P seismic phase and the S seismic phase in each segmented waveform; Step 1.3: Analyze and judge the probability values of the P seismic phase and the S seismic phase in each segmented waveform: If the probability values of the P seismic phase and the S seismic phase of a certain segmented waveform are both greater than the segmented preset threshold, it is determined that the current segmented waveform is effective seismic phase data; otherwise, it is determined that there is no effective seismic phase data in the current segmented waveform; Step 1.4: Repeat Step 1.3 until all segmented waveforms of each seismograph station have completed the analysis and judgment, and respectively output all the effective seismic phase data of each seismograph station, and the effective seismic phase data of each seismograph station are arranged in chronological order.

3. The template matching earthquake detection method based on artificial intelligence according to claim 2, wherein, In Step 2, the specific steps of correlating the extracted effective seismic phase data of each station to determine the effective seismic waveforms among them are as follows: Step 2.1: Obtain the seismograph station coordinates of the seismograph point corresponding to each effective seismic phase data and the acquisition time of each effective seismic phase data, and then set a source-limiting circle for limiting the possible range of the earthquake source, an extraction time interval for delimiting the time range of data extraction, and the effective waveform intercepting time length; Step 2.2: Allocate the effective seismic phase data of each seismograph station to the corresponding extraction time intervals where the corresponding moments are arranged in chronological order of acquisition time, and then form a seismic phase data group with the effective seismic phase data of each seismograph station within each extraction time interval; Step 2.3: Take out a seismic phase data group in sequence according to the arranged time order, and decompose the P seismic phase wave and S seismic phase wave in each effective seismic phase data in the current seismic phase data group; Step 2.4: Determine whether there is still unextracted effective seismic phase data in the seismic phase data group. If there is still unextracted effective seismic phase data, take out an effective seismic phase data and enter Step 2.5, otherwise enter Step 2.6; Step 2.5: Judge the correlation of the P seismic phase wave and S seismic phase wave in the taken-out effective seismic phase data: If the arrival time of the S seismic phase wave in the effective seismic phase data of a certain seismograph station is greater than the arrival time of the P seismic phase wave, determine that the current effective seismic phase data is possible event data, and then return to Step 2.

4. Otherwise, determine that the current effective seismic phase data is invalid event data, delete the current effective seismic phase data, and then return to Step 2.4; Step 2.6: Judge whether the number of possible event data in the current seismic phase data group is greater than the event number threshold. If the number is less than the event number threshold, determine that the current seismic phase data group is an invalid data group, and then enter Step 2.

10. Otherwise, enter Step 2.7; Step 2.7: Calculate the distance D between each seismograph station and the earthquake source respectively according to the arrival time of the S seismic phase wave, the arrival time of the P seismic phase wave, the S seismic phase wave velocity and the P seismic phase wave velocity in each possible event data. Then, draw each earthquake source location circle with the seismograph point coordinates of each seismograph station as the center and the corresponding distance D as the radius. After drawing the earthquake source location circles corresponding to each possible event data, mark the intersection points of each earthquake source location circle as the positioning reference points; Step 2.8: Select each positioning reference point as the center of the earthquake source limiting circle in sequence, then determine the number of points of the positioning reference points within each earthquake source limiting circle respectively, and then compare the determined numbers of points to find the maximum number of points. If the maximum number of points is greater than the point number threshold, determine that the current seismic phase data group is an effective data group, and then enter Step 2.

9. Otherwise, determine that the current seismic phase data group is an invalid data group, and then enter Step 2.10; Step 2.9: Find the earthquake source limiting circle corresponding to the maximum number of points, then determine each positioning reference point within the earthquake source limiting circle, then find the corresponding earthquake source location circle according to each positioning reference point, then determine each seismograph station according to each earthquake source location circle, and finally intercept the effective earthquake waveform from the effective seismic phase data of each seismograph station according to the effective waveform intercept time length, and the effective earthquake waveform includes the complete waveforms of the S seismic phase wave and the P seismic phase wave; Step 2.10: Judge whether there is still an unextracted seismic phase data group. If there is still an unextracted seismic phase data group, return to Step 2.

3. Otherwise, output the effective earthquake waveforms of each seismograph station.

4. The template matching earthquake detection method based on artificial intelligence according to claim 1, wherein In Step 3, the specific steps of intercepting the artificial earthquake monitoring waveform according to the earthquake events of each seismograph station are as follows: Step 3.1: Determine the seismic events of each seismograph station according to the identified artificial earthquake catalog, and obtain the artificial monitoring data of each seismic event; Step 3.2: Use the TAUP ray tracing technique to calculate the arrival times of the S-wave phase and the P-wave phase of each seismic event to each seismograph station; Step 3.3: Set the start point of the effective waveform truncation interval to 7 seconds before the arrival time of the P-wave phase, then set the end point of the effective waveform truncation interval to 15 seconds after the arrival time of the P-wave phase, and then extract the corresponding artificial earthquake monitoring waveforms from the artificial monitoring data of each seismograph station according to the set effective waveform truncation interval.

5. The method for template matching earthquake detection based on artificial intelligence according to claim 1, characterized in that, In Step 4, the specific steps for constructing the template seismic waveform library by combining the machine-detected seismic waveform templates and the manually-detected seismic waveform templates are as follows: Step 4.1: Collect each machine-detected seismic waveform template and each manually-detected seismic waveform template, and obtain the seismograph stations corresponding to the machine-detected seismic waveform templates and the seismograph stations corresponding to the manually-detected seismic waveform templates; Step 4.2: Use the names of each seismograph station as the first-level directories of the template seismic waveform library, and place each machine-detected seismic waveform template and each manually-detected seismic waveform template under the corresponding first-level directory; Step 4.3: Extract the P-wave phase from the machine-detected seismic waveform templates or the manually-detected seismic waveform templates under each first-level directory, classify the obtained seismic waveform templates according to the amplitude of the P-wave phase, and use the classification level as the second-level directory. Then place each machine-detected seismic waveform template and each manually-detected seismic waveform template under the corresponding second-level directory. If there are multiple seismic waveform templates under a certain level, delete the redundant seismic waveform templates so that only one seismic waveform template is retained for each level, and finally form the template seismic waveform library.

6. The method for template matching earthquake detection based on artificial intelligence according to claim 5, wherein In Step 4.3, the specific steps for classifying the obtained seismic waveform templates according to the amplitude of the P-wave phase are as follows: First, decompose each machine-detected seismic waveform template and each manually-detected seismic waveform template collected in Step 4.1 to obtain the P-wave phase in each seismic waveform template; Then, screen out the maximum amplitude value in each P-wave phase and define it as Tmax, and screen out the minimum amplitude value in each P-wave phase and define it as Tmin; Then set 0.8Tmin as the minimum level value, set 1.6Tmax as the maximum level value, and equally spacedly set 8 level values between 0.8Tmin and 1.6Tmax. Use the minimum level value, the maximum level value, and the 8 level values in between to divide the amplitude of the P-wave phase into 10 level intervals; Finally, determine which level in the 10 level intervals the amplitude of the P-wave phase belongs to according to the amplitude of the P-wave phase, which is the classification level of the corresponding seismic waveform template.

7. The method for template matching earthquake detection based on artificial intelligence according to claim 5, characterized in that, In Step 5, the specific steps for truncating each seismic waveform template in the template seismic waveform library to obtain the template waveforms of each seismic phase are as follows: Step 5.1: Obtain the seismic waveform template in a secondary directory of the template seismic waveform library, and separate the P-phase wave and the S-phase wave from the obtained seismic waveform template; Step 5.2: Perform separate truncations on the separated P-phase wave and S-phase wave: Take the waveform in the interval from the first 1 s to the last 5 s of the P-phase wave as the P-phase wave template waveform, and take the waveform in the interval from the first 1 s to the last 5 s of the S-phase wave as the S-phase wave template waveform; Step 5.3: Replace the seismic waveform template in the corresponding secondary directory with the obtained P-phase wave template waveform and S-phase wave template waveform, and then determine whether there is still a seismic waveform template in the template seismic waveform library that has not been truncated. If there is still a seismic waveform template that has not been truncated, return to Step 5.1; otherwise, proceed to Step 5.4; Step 5.4: Update the template seismic waveform library: Name the P-phase wave template waveforms in each secondary directory as P waves, and name the S-phase wave template waveforms in each secondary directory as S waves.

8. The method for template matching earthquake detection based on artificial intelligence according to claim 1, wherein In Step 6, the specific steps for truncating the continuous waveform data to obtain the continuous waveform to be inspected suspected of having a seismic event are as follows: Step 6.1.1: Extract each data segment with a waveform in the continuous waveform data, and then perform waveform analysis on each extracted data segment to obtain the suspected waveforms of each data segment; Step 6.1.2: Use the deep learning detection model to identify the separated suspected waveforms, and determine whether each suspected waveform is a P-phase waveform or an S-phase waveform. If P-phase waveforms and S-phase waveforms are identified, proceed to Step 6.1.3; otherwise, determine that there is no seismic event in the continuous waveform data, and then proceed to Step 6.1.6; Step 6.1.3: Perform statistical judgment on the identified P-phase waveforms and S-phase waveforms. If there is one P-phase waveform and one S-phase waveform, proceed to Step 6.1.4; if there are multiple P-phase waveforms and S-phase waveforms, proceed to Step 6.1.5; Step 6.1.4: Perform waveform matching on the S-phase wave and the P-phase wave. If the S-phase wave and the P-phase wave belong to the same seismic source and the arrival time of the S-phase wave is greater than the arrival time of the P-phase wave, determine that there is a suspected seismic event in the continuous waveform data. Then, splice the S-phase wave and the P-phase wave in chronological order according to the arrival time of the S-phase wave and the arrival time of the P-phase wave to obtain the continuous waveform to be inspected, and then proceed to Step 6.1.6; otherwise, determine that there is no seismic event in the continuous waveform data, and then proceed to Step 6.1.6; Step 6.1.5: Perform waveform matching on each P-phase waveform and each S-phase waveform, and use the matched P-phase waveform and S-phase waveform as a phase matching group. If the number of obtained phase matching groups ≥ 1, determine that there is a suspected seismic event in the continuous waveform data. Then, splice the S-phase wave and the P-phase wave in each phase matching group in chronological order according to the arrival time of the S-phase wave and the arrival time of the P-phase wave to obtain each continuous waveform to be inspected, and then proceed to Step 6.1.6; otherwise, determine that there is no seismic event in the continuous waveform data, and then proceed to Step 6.1.6; Step 6.1.6, output the determination result: If it is determined that there is a suspected earthquake event in the continuous waveform data, output each continuous waveform to be detected obtained by splicing; if it is determined that there is no earthquake event in the continuous waveform data, output the result of not obtaining the continuous waveform to be detected.

9. The method for template matching earthquake detection based on artificial intelligence according to claim 8, characterized in that, In Step 6.1.5, the specific steps for waveform matching of each P-phase waveform and each S-phase waveform are as follows: Step a, perform filtering processing on each P-phase waveform and each S-phase waveform to obtain the denoised P-phase waveform and S-phase waveform; Step b, arrange each S-phase waveform in ascending order of arrival time to form an S-wave sequence, and arrange each P-phase waveform in ascending order of arrival time to form a P-wave sequence; Step c, take out the S-phase waveform with the smallest arrival time from the S-wave sequence, and then find out each P-phase waveform in the P-wave sequence whose arrival time is smaller than the taken-out S-phase waveform. If no corresponding P-phase waveform is found, go to Step h; Step d, obtain the maximum amplitude of the taken-out S-phase waveform as Hs and the maximum amplitude of each of the found P-phase waveforms as Hp, and then set the height of the intercept line to 20%Hs. Use the intercept line with a height of 20%Hs to perform waveform interception on the taken-out S-phase waveform and the found P-phase waveforms, obtain the distribution time width ΔS of the S-phase waveform above the intercept line and the distribution time width ΔP of each P-phase waveform, and then count the number of times Ns that the S-phase waveform crosses the intercept line up and down within the range of the distribution time width ΔS and the number of times Np that the corresponding P-phase waveform crosses the intercept line up and down within each range of the distribution time width ΔP; Step e, find out the P-phase waveforms with Hp < Hs among each P-phase waveform, and use the found P-phase waveforms as the P-phase waveforms after the first-level screening. If no corresponding P-phase waveform is found, go to Step h; Step f, find out each P-phase waveform with 0.85 < Np / Ns < 1.15 among the P-phase waveforms after the first-level screening, and use the found P-phase waveforms as the P-phase waveforms after the second-level screening. If no corresponding P-phase waveform is found, go to Step h; Step g, find out the distribution time width ΔP with the smallest difference from the distribution time width ΔS among the P-phase waveforms after the second-level screening, and determine the P-phase waveform corresponding to the found distribution time width ΔP as the one that matches the current S-phase waveform, and delete the found P-phase waveform from the P-wave sequence, and then go to Step h; Step h, determine whether there is still an S-phase waveform in the S-wave sequence. If there is still an S-phase waveform, return to Step c; otherwise, go to Step i; Step i, output the paired P-phase waveforms and S-phase waveforms that match.

10. The template matching earthquake detection method based on artificial intelligence according to claim 7, wherein In Step 6, the specific steps for performing matching calculations between the continuous waveform to be detected and the template waveforms of each seismic phase of each seismic waveform template in the template seismic waveform library are as follows: Step 6.2.1, determine the seismograph station to which the continuous waveform to be detected belongs according to the data source of the continuous waveform to be detected, and then determine the first-level directory in the template seismic waveform library according to the seismograph station; Step 6.2.2, calculate the maximum amplitude of the P phase waveform in the continuous waveform to be detected, and then match the corresponding grade classification according to the maximum amplitude of the P phase waveform, so as to determine the secondary directory under the primary directory, and then determine whether there are P waves and S waves in the determined secondary directory. If there are P waves and S waves, go to step 6.2.3, otherwise go to step 6.2.4; Step 6.2.3, determine the maximum amplitude S1 of the S seismic phase waveform in the continuous waveform to be tested and the maximum amplitude S2 of the S wave in the secondary directory, and then calculate the ratio of the maximum amplitude S1 to the maximum amplitude S2. If the ratio is within the range of 0.9 to 1.1, it is determined that the current continuous waveform to be tested is matched successfully, otherwise it is determined that the current continuous waveform to be tested fails to match, and then proceed to step 6.2.6; Step 6.2.4, searching for P waves and S waves in the secondary directories of adjacent levels step by step, and after finding the P waves and S waves in the secondary directories of adjacent levels, proceed to step 6.2.5; Step 6.2.5, determine the maximum amplitude P1 of the P seismic phase waveform and the maximum amplitude S1 of the S seismic phase waveform in the continuous waveform to be tested, then determine the maximum amplitude P2 of the P wave and the maximum amplitude S2 of the S wave in the secondary directory of the adjacent level, then calculate the matching degree G as P1×S2 / P2×S1, and then determine whether the matching degree G is within the matching threshold range of 0.8~1.

2. If it is within the matching threshold range, it is determined that the current continuous waveform to be tested is matched successfully, and the P seismic phase waveform and the S seismic phase waveform in the continuous waveform to be tested are stored as P waves and S waves in the secondary directory matching the P seismic phase waveform level in the template seismic waveform library, otherwise it is determined that the current continuous waveform to be tested fails to match, and then enter step 6.2.6; Step 6.2.6, output the matching calculation results.

Images