Seismic exploration auxiliary data quality control method and application thereof

Abstract

The invention belongs to the technical field of petroleum geophysical prospecting acquisition operation, and discloses a seismic prospecting auxiliary data quality control method and application thereof, the method performs quality control on a seismic source vehicle work log file, an R file, an S file and an X file returned by field production, and the file accuracy after quality control is relatively high. After excitation point seismic source index quality control is completed, the method is used for carrying out early warning on the seismic source state through LSTM, and the application effect in seismic source early warning is ideal. The method is suitable for petroleum geophysical prospecting collection operation data processing.

Description

Technical Field

[0001] The present invention belongs to the technical field of oil geophysical prospecting acquisition operations, and relates to a method for quality control of seismic exploration auxiliary data, specifically a method for quality control of seismic exploration auxiliary data and its application. Background Art

[0002] Onshore seismic exploration auxiliary data (SPS, shell processing support format for land 3D surveys), also known as SPS files, can enable the seismic data processing center to load the observation system for seismic data through the SPS files and restore the geometric relationship between the shot points and the receiving points during field seismic exploration. The SPS files include a shot point file (S file), a receiving point file (R file), and a relationship file (X file) describing the arrangement relationship between the shot points and the receiving points.

[0003] For the implementation of onshore seismic exploration projects, the SPS files are required to be accurate. However, during actual field production, the geophones at the receiving points may move due to oil pipe construction, road construction, damage to the geophones, etc., which is also called offset, resulting in multiple records with different point indexes and point codes at the same receiving point; or due to the failure to return the starting vibration time of the vibration source vehicle, the working state indicators exceeding the contract standards, etc., the vibration source at the shot point is often excited multiple times or the vibration source at the shot point is unqualified in six vibration source working indicators including peak distortion, average distortion, peak phase, average phase, peak output, and average output, resulting in errors in the SPS files returned from field production. Usually, they need to pass quality control to be qualified before they can be used.

[0004] Currently, the quality control methods for SPS files usually only focus on the quality control of the S file and the X file. For example, the method of Yao Hongliang (Automatic sorting of SPS auxiliary data during the efficient acquisition of the vibration source of G3i instruments, Geophysical Prospecting Equipment, Vol. 29, No. 1). Due to ignoring the importance of the quality control of the R file, the accuracy of the quality control files obtained is relatively low. In addition, the method for quality control using 3D GIS disclosed in the Chinese patent application with the publication number CN104280764A also has the problem that it cannot check the condition of the vibration source at the shot point. With the development of seismic exploration technology, the requirements for the accuracy and efficiency of the quality control of SPS files are getting higher and higher. Summary of the Invention

[0005] To solve the above-mentioned deficiencies in the prior art, the present invention aims to provide a method for quality control of seismic exploration auxiliary data, so as to achieve the purpose of improving the accuracy and efficiency of the quality control of SPS files;

[0006] The present invention also provides an application of a seismic exploration auxiliary data quality control method in the warning of the source state, so as to achieve the purpose of accurate warning.

[0007] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0008] A seismic exploration auxiliary data quality control method, which imports the source vehicle work log file, R file, S file, X file and seismic acquisition instrument log file returned from field production into the database, and sequentially performs the following steps in the database:

[0009] S1. Preliminary quality control of the R file, including the collation of the R file returned from field production and coordinate replacement;

[0010] S2. Preliminary quality control of the S file, including the collation of the S file returned from field production;

[0011] S3. Preliminary quality control of the X file, including the modification and collation of the X file returned from field production;

[0012] S4. Repeated shot quality control, including checking for repeated shots in the files obtained in S2 and S3;

[0013] S5. Quality control of the source point source index, including checking the source point source index of the shot points in the files obtained in S2 and S3; obtaining the S file, X file and R file after quality control, and extracting the qualified shot point records in the source vehicle work log file according to the line number and point number of the shot points in the S file after quality control, to obtain the source vehicle work log file after quality control, that is, the seismic exploration auxiliary data quality control is completed.

[0014] As a limitation of the present invention, in step S1, the collation of the R file returned from field production is to group the records in the R file returned from field production according to the line number and point number of the receiving points. When the number of records in the group is ≥ 2, the record with the highest priority is retained, denoted as OffsetFile, and the records not to be retained are deleted to obtain the collated R file;

[0015] The coordinate replacement is to replace the coordinates of the receiving points in the collated R file with the actually measured coordinates to obtain the R file after coordinate replacement;

[0016] The record with the highest priority is the first normal record of the receiving point;

[0017] The retention of the record with the highest priority is to retain one record according to the following priority from high to low: 1G1, 1KL, nG1, n > 1 and n ∈ N*, where G1 indicates that the record is normal, n indicates the number of changes in the point position information of the receiving point, and KL indicates that the record is abnormal;

[0018] Compare the line numbers and point numbers of the receiving points in the R file after coordinate replacement with those of the receiving points in the X file returned by the field production. When there are missing receiving point records in the R file after coordinate replacement, manually check and then export the correct R file again, import it into the database, and then execute all the steps starting from S1 until there are no missing receiving point records in the R file after coordinate replacement, thus completing the preliminary quality control of the R file.

[0019] As a further limitation of the present invention, the preliminary quality control of the S file is to group the records in the S file returned by the field production according to the line numbers and point numbers of the shot points in the S file returned by the field production, retain the shot point records with the largest point index within the group, delete the records that are not retained, obtain the retained S file, calculate the coordinate distance of the shot points with the same line numbers and point numbers in the retained S file and the S file of the task book, and when the distance exceeds the range allowed by the task book, delete the shot point from the retained S file, thus completing the preliminary quality control of the S file.

[0020] As a further limitation of the present invention, when any shot point in the S file after preliminary quality control is fired after the receiving point time in the OffsetFile and the line number of the receiving point in the OffsetFile is required, the X file returned by the field production corresponding to such shot points in the S file needs to be modified;

[0021] The modification of the X file returned by the field production corresponding to such shot points, that is, the modification of the X file returned by the field production, is to extract the line number from the OffsetFile, denoted as Line, extract all the records containing this line number from the R file after preliminary quality control according to Line, denoted as ALLR, and replace the records in ALLR with duplicate line numbers and point numbers in the OffsetFile with the records in the OffsetFile. Then, increment the point index of ALLR after replacement by 1 and add it to the R file after preliminary quality control. At this time, for the records with the line number Line, there are two types, one is the record before rectification, denoted as A, and the other is the record after rectification, denoted as B; modify the receiving point index of the records containing Line in the X file returned by the field production to the point index of B to obtain the modified X file.

[0022] When any shot point in the S file after preliminary quality control is fired before the receiving point time in the OffsetFile or the line number of the receiving point in the OffsetFile is not required, there is no need to modify the X file returned by the field production.

[0023] As a further limitation of the present invention, the X file sorting extracts the corresponding records in the X file according to the line number, point number, and point index of the excitation points in the S file after preliminary quality control, denoted as tempX;

[0024] The X file refers to the modified X file or the X file returned from field production that does not need to be modified;

[0025] Extract the corresponding records from the X file in the task book according to the line number and point number of the excitation points in the S file after preliminary quality control, and store them in the temporary data table together with tempX, denoted as temp;

[0026] Calculate the number of identical records in temp according to the shot point line number, shot point number, receiving line number, starting point number of the receiving line, and ending point number of the receiving line. When the number of identical records is not 2, conduct manual inspection and import the correct file into the database;

[0027] Start executing all steps from S1 until the number of identical records in temp is 2 for all, that is, the preliminary quality control of the X file is completed.

[0028] As a further limitation of the present invention, the duplicate shot quality control extracts the relationship records from the X files that have completed quality control in the past according to the line number and point number of the excitation points in the S file after preliminary quality control. When the extracted relationship file record is 0, it indicates that this shot is not a duplicate shot;

[0029] When the extracted relationship file record is not 0 and is consistent with the relationship file record of this shot in tempX, this shot is a duplicate shot. Delete this shot point from the S file and X file after preliminary quality control, that is, the duplicate shot quality control is completed.

[0030] As a further limitation of the present invention, the excitation point seismic source index quality control extracts the file number from the X file after preliminary quality control according to the line number, point number, and point index of the excitation points extracted from the S file after preliminary quality control, and then extracts six seismic source working indicators of peak distortion, average distortion, peak phase, average phase, peak output, and average output from the seismic source vehicle work log file returned from field production according to the file number. When the seismic source working indicators are unqualified, check whether this excitation point has been excited multiple times. If it has been excited only once, directly delete it from the S file and X file after preliminary quality control;

[0031] If it is excited more than twice, all records are extracted from the source vehicle work log file according to the line number and point number of the excitation point, the excitation points with qualified source work indicators are selected, the point index of this excitation point is denoted as N, the point index of this excitation point in the S file after preliminary quality control is modified to N, and the point index of the excitation point in the relationship record of this excitation point in the X file after preliminary quality control is also modified to N; if no qualified excitation point can be selected, this excitation point is deleted from the S file after preliminary quality control, that is, the quality control of the source indicators of the excitation point is completed.

[0032] The present invention also provides an application of a seismic exploration auxiliary data quality control method in the warning of the source state. After the quality control of the source indicators of the excitation point is completed, LSTM is used to warn the source state, including the following steps:

[0033] P1. Extract all records of the source under study during the study time from the source vehicle work log file after quality control, sort them in ascending order of time, extract the data corresponding to the source state indicators that need to be warned, denoted as Record, perform maximum-minimum normalization on Record by column, the number of columns in Record is denoted as C, and the number of rows is denoted as M, to obtain the standardized data;

[0034] P2. Divide the standardized data into M training samples, and divide the training samples into a training set and a test set. The division ratio of the training set to the test set is (1 - RTT): RTT. The training set contains (1 - RTT) × M training samples, and the test set contains RTT × M training samples, where the range of RTT is (0, 1);

[0035] P3. Apply LSTM to the training samples. At this time, the number of LSTM layers is denoted as Layer, where Layer > 0 and is a positive integer; the number of LSTM neurons in the i-th layer is denoted as j, where i is any integer from 1 to Layer, and j > 0 and is a positive integer; the neuron inactivation rate between adjacent layers is denoted as DeadR;

[0036] Construct a source warning model according to the structure of the input layer, LSTM layers with different numbers of neurons in Layer layers, a Dropout layer with inactivation rate of DeadR, a ReLu activation function connecting the last LSTM layer and the output layer, and the output layer.

[0037] P4. Use the training set divided in P2 to train the earthquake source warning model, and use an optimizer to optimize the entire earthquake source warning model. When optimizing, use the mean absolute error as the loss function. Each time during training, input BZ training samples. When all the training samples in the training set have been traversed once, it is one training cycle. When training for Epoch cycles and the value of the loss function < 0.005, the training of this model ends. Subsequently, use the samples in the test set to test the earthquake source warning model, and calculate the absolute errors of the peak distortion and the average distortion between the output value of the earthquake source warning model and the expected value of the test set. When the average absolute errors of both are < 5%, stop training; otherwise, train again until the average absolute errors of both are < 5%.

[0038] The Epoch > 0 and is a positive integer;

[0039] The BZ is in the range of [1, RTT×M] and is a positive integer.

[0040] As a limitation of the present invention, the ReLu activation function compares the input value with 0 and takes the maximum value of the two;

[0041] The optimizer is the adaptive moment estimation optimizer, and the adaptive moment estimation optimizer is the "Adam" optimizer.

[0042] The present invention also provides a computer-readable storage medium, which stores a computer program for executing the seismic exploration auxiliary data quality control method described in any one of the above technical solutions.

[0043] Due to the adoption of the above technical solutions, compared with the prior art, the beneficial effects obtained by the present invention are:

[0044] (1) By performing preliminary quality control on the R file, S file, X file, repeated shot quality control, and excitation point seismic source index quality control on the files returned from field production, and on this basis, obtaining the seismic source vehicle workday log file after quality control, the seismic exploration auxiliary data quality control is completed. Applying the data after quality control to the earthquake source state warning has a significant effect, indicating that the accuracy of this seismic exploration auxiliary data quality control method is relatively high.

[0045] (2) In the present invention, for the preliminary quality control of the R file, records with different point indexes and point codes that appear at the same receiving point due to offsets are reasonably deleted, and only one receiving point record is retained; for the preliminary quality control of the S file, records of multiple excitations at the excitation point are reasonably deleted, and only one excitation point record is retained; for the preliminary quality control of the X file, through the investigation of relationship records, the integrity of the entire field production return file is further ensured; for the repeated shot quality control and the excitation point source index quality control, repeated shots and unqualified sources are investigated, and further quality control is performed on the S file and the X file; the quality control that systematically combines the preliminary quality control of the R file, the preliminary quality control of the S file, the preliminary quality control of the X file, the repeated shot quality control, and the excitation point source index quality control has high accuracy and efficiency.

[0046] The present invention is applicable to the technology of petroleum geophysical prospecting acquisition operations, and is used for quality control of SPS files and source vehicle work log files returned from field production, and for completing source warning operations based on this. Brief Description of the Drawings

[0047] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments.

[0048] Figure 1 It is a schematic diagram of the modification process of the X file returned from field production in Embodiment 1 of the present invention;

[0049] Figure 2 It is a schematic diagram of the excitation point source index quality control process in Embodiment 1 of the present invention;

[0050] Figure 3 It is a schematic diagram of the source warning step process in Embodiment 2 of the present invention;

[0051] Figure 4 It is a graph of the change of the loss function during the training process in Embodiment 2 of the present invention;

[0052] Figure 5 It is a statistical graph of the absolute error of the peak distortion between the output value of the source warning model and the test set in the test results of the source warning model in Embodiment 2 of the present invention;

[0053] Figure 6 It is a statistical graph of the absolute error of the average distortion between the output value of the source warning model and the test set in the test results of the source warning model in Embodiment 2 of the present invention. Detailed Description of the Specific Embodiment

[0054] The present invention will be further described in detail below through specific embodiments and drawings. It should be understood that the described embodiments are only used to explain the present invention and do not limit the present invention.

[0055] Embodiment 1 A Method for Quality Control of Seismic Exploration Auxiliary Data

[0056] In a 3D seismic exploration project in the Middle East, a super-large digital acquisition system with over 200,000 channels is used to import the source vehicle work log files, R files, S files, X files, and seismic acquisition instrument log files returned from daily field production into the database. Among them, the number of records in the X files returned from the field each day exceeds 350,000. The following steps are carried out in sequence in the database:

[0057] S1. Use the corresponding Structured Query Language (SQL) of the database to extract the line numbers and point numbers of the receiving points in the R files returned from field production. Group the records in the R files returned from field production according to the line numbers and point numbers. When the number of records in the group is ≥2, it indicates that the receiving point corresponding to the line number and point number has been rectified. At this time, retain one record according to the following priority from high to low: 1G1, 1KL, nG1, n>1 and n∈N*, denoted as OffsetFile. Use the SQL statement to delete the un-retained records to obtain the sorted R file;

[0058] Replace the coordinates of the receiving points in the sorted R file with the actually measured coordinates to obtain the R file with coordinate replacement;

[0059] Compare the line numbers and point numbers of the receiving points in the R file with coordinate replacement and the X files returned from field production. When there are missing receiving point records in the R file with coordinate replacement, the staff needs to conduct a check, and at the same time contact the front-line staff. After the check, export the correct R file again, import it into the database, and start executing all the steps from S1 until there are no missing receiving point records in the R file with coordinate replacement, that is, complete the preliminary quality control of the R file.

[0060] S2. Use SQL to extract the line numbers and point numbers of the shot points in the S files returned from field production. Group the records in the S files returned from field production according to the line numbers and point numbers. When there are multiple records for the same shot point, specifically reflected in the point index, when shooting multiple times, the index of the first shot is "1", the index of the second shot is "2", and so on. When sorting, by default, the last shot of a certain shot point is "qualified", retain the record of the shot point with the largest point index in the group, delete the un-retained records to obtain the retained S file. Calculate the coordinate distance of the shot points with the same line numbers and point numbers in the retained S file and the S file in the task book. When the distance exceeds the range allowed by the task book, delete the shot point from the retained S file to complete the preliminary quality control of the S file.

[0061] S3. Record the time of the receiving point in OffsetFile as T, use SQL to extract the line number from OffsetFile and record it as Line. According to Line, extract all records containing this line number from the R file after preliminary quality control and record them as ALLR. Replace the records in ALLR that have duplicate line numbers and point numbers with the records in OffsetFile. Then increment the point index of ALLR after replacement by 1. For example, if the point index in all records corresponding to a certain line number is "2", then it becomes "3" after incrementing the point index. Finally, add ALLR with the modified point index to the R file after preliminary quality control. At this time, for the records with line number Line, there are two types: one is the record before rectification, denoted as A, and the other is the record after rectification, denoted as B;

[0062] Use SQL to extract the firing time of the firing point in the S file after preliminary quality control and compare it with T to obtain the firing points that are fired at T and later. According to the line number and point number of this firing point, extract the records in the X file returned by field production. If the receiving point line number in this record contains Line, modify the receiving point index of the record in the X file returned by field production corresponding to this firing point and containing Line to the point index of B to obtain the modified X file. The step flow is shown in Figure 1 。

[0063] Use SQL to extract the line number, point number, and point index of the firing point in the S file after preliminary quality control to extract the corresponding records in the modified X file and record them as tempX;

[0064] Then, according to the line number and point number of the firing point in the S file after preliminary quality control, extract the corresponding records from the X file in the task book and store them in the temporary data table together with tempX, denoted as temp;

[0065] Calculate the number of identical records in temp according to the shot point line number, shot point number, receiving line number, starting point number of the receiving line, and ending point number of the receiving line. If there are no problems with the records in tempX, the number of identical records in temp should all be 2; if there are problems, the staff needs to conduct a check, contact the front-line staff at the same time, export the correct file after the check, import it into the database, and then execute all steps from S1 until the number of identical records is 2 and there are no problems with the records in tempX, completing the preliminary quality control of the X file.

[0066] S4. Use SQL to extract the line number and point number of the firing point in the S file after preliminary quality control. According to the line number and point number, extract the relationship records from the X file that has completed quality control in the past. When the number of extracted relationship file records is 0, this shot is not a duplicate shot;

[0067] When the extracted relationship file record is not 0 and is consistent with the relationship file record of this gun in tempX, this gun is a duplicate gun. Delete this gun point from the S file and X file after preliminary quality control to complete the duplicate gun quality control.

[0068] S5. Use SQL to extract the line number, point number, and point index of the shot points from the S file after preliminary quality control. Based on the line number, point number, and point index, extract the file number from the X file after preliminary quality control. Then, based on the file number, extract six vibration source working indicators, namely peak distortion, average distortion, peak phase, average phase, peak output, and average output, from the vibration source vehicle work log file returned from field production. When the vibration source working indicators are unqualified, check whether this shot point has been fired multiple times. If it has been fired only once and at this time the point index is 1, then directly delete it from the S file and X file after preliminary quality control.

[0069] If it has been fired more than twice and at this time the point index > 1, then based on the line number and point number of this shot point, extract all the records from the vibration source vehicle work log file returned from field production, select the shot points with qualified vibration source working indicators, record the point index of this shot point as N, modify the point index of this shot point in the S file after preliminary quality control to N, and also modify the point index of the shot point in the relationship record of this shot point in the X file after preliminary quality control to N. If no qualified shot points can be selected, then delete this shot point from the S file after preliminary quality control to complete the quality control of the shot point vibration source indicators. The step process is shown in Figure 2 。

[0070] Finally, obtain the S file, X file, and R file after quality control. Extract the qualified shot point records from the vibration source vehicle work log file returned from field production based on the line number and point number of the shot points in the S file after quality control to obtain the vibration source vehicle work log file after quality control, and complete the quality control of seismic exploration auxiliary data. Refer to "Algorithm Illustration" and use the big O notation to characterize the efficiency of this method. The efficiency of the seismic exploration auxiliary data quality control method of the present invention is O(350,000).

[0071] Comparative Example 1 Traditional Quality Control Method

[0072] In a 3D seismic exploration project in the Middle East, a super-large digital acquisition system with over 200,000 channels was used. The number of X file records returned from field production every day exceeded 350,000. The X file returned from field production was checked in a double-loop manner, that is, a single record was sequentially extracted from the X file returned from field production and then checked against each record in the task book. Refer to "Algorithm Illustration" and use the big O notation to characterize the efficiency of this method. The efficiency of the traditional quality control method is O(122.5 billion).

[0073] From an algorithmic perspective, the O(350,000) algorithm is faster, and the seismic exploration auxiliary data quality control method of the present invention has higher efficiency.

[0074] Example 2 Application of a seismic exploration auxiliary data quality control method in the early warning of the source state

[0075] Based on Example 1, LSTM is used to give an early warning of the source state. The specific steps are as follows. The step process is shown in Figure 3 :

[0076] P1. Extract all the records of a certain source from the start time to the end time from the source vehicle work log file after quality control, sort them in ascending order of time, and extract the data corresponding to the source state indicators that need to be warned, denoted as Record. Perform maximum-minimum normalization on Record by column. The number of columns in Record is denoted as 2, and the number of rows is denoted as 86917 to obtain the standardized data.

[0077] P2. Divide the standardized data into 86917 training samples, and divide the training samples into a training set and a test set. The division ratio of the training set to the test set is 0.95:0.05. The training set contains 82571 training samples, and the test set contains 4346 training samples.

[0078] P3. Apply LSTM to the training samples. The number of LSTM layers is 3. The neuron inactivation rate of the first layer is 0.3, and the neuron inactivation rates of the second and third layers are both 0.5.

[0079] According to the structure of the input layer, the first LSTM layer with 2048 neurons, the Dropout layer with 0.3 inactivation, the second LSTM layer with 1024 neurons, the Dropout layer with 0.5 inactivation, the third LSTM layer with 512 neurons, the Dropout layer with 0.5 inactivation, the ReLu activation function, and the output layer, construct the source warning model.

[0080] P4. Use the training set divided in P2 to train the source warning model. The "Adam" optimizer is used to optimize the entire model. The mean absolute error is used as the loss function during optimization. 64 training samples are input each time. When all the training samples in the training set are traversed once, it is recorded as one training cycle. When training for 580 cycles, see Figure 4 , the mean absolute error is stable at about 0.004, and the training of this model ends. The "burr" phenomenon in the curve is caused by the "Dropout" layer in the model.

[0081] Subsequently, samples in the test set were used to test the earthquake source warning model. After obtaining the test results, the absolute error between the two was calculated, and the results are statistically shown as Figure 5 , Figure 6 shown. For peak distortion, most of the absolute errors are less than 0.053. Calculate the absolute error of each sample, sum them up, and then divide by the number of samples. The average absolute error is 2.3%. For average distortion, most of the absolute errors are less than 0.05. Calculate the absolute error of each sample, sum them up, and then divide by the number of samples. The average absolute error is 1.8%. Through testing, it is considered that this model has the warning ability.

[0082] Embodiment 3 A computer-readable storage medium

[0083] This embodiment provides a computer-readable storage medium that stores a computer program. When the computer program is executed by a processor, it implements the seismic exploration auxiliary data quality control method of Embodiment 1.

[0084] On this computer-readable storage medium, there are non-temporary computer-readable instructions stored. When these non-temporary computer-readable instructions are run by a processor, all or part of the steps of the methods of the foregoing embodiments are executed.

[0085] The above-mentioned computer-readable storage medium includes but is not limited to: optical storage media (such as CD-ROM and DVD), magneto-optical storage media (such as MO), magnetic storage media (such as magnetic tapes or external hard drives), media with built-in rewritable non-volatile memory (such as memory cards), and media with built-in ROM (such as ROM cartridges).

[0086] It should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still modify the technical solutions recorded in the above embodiments or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. An auxiliary data quality control method for seismic exploration, characterized in that, It imports the source vehicle work log files, R files, S files, X files, and seismic acquisition instrument log files returned from field production into the database, and then sequentially performs the following steps in the database: S1. Preliminary quality control of R files, including sorting the R files returned from field production and coordinate replacement; S2. Preliminary quality control of S files, including sorting the S files returned from field production; S3. Preliminary quality control of X files, including modifying and sorting the X files returned from field production; S4. Duplicate shot quality control, including checking for duplicate shots in the files obtained from S2 and S3; S5. Quality control of source point source indicators, including checking the source point source indicators for the files obtained from S2 and S3; obtaining the S files, X files, and R files after quality control, and extracting the qualified source point records in the source vehicle work log file according to the line numbers and point numbers of the source points in the S file after quality control, to obtain the source vehicle work log file after quality control, that is, completing the quality control of seismic exploration auxiliary data.

2. The seismic exploration auxiliary data quality control method according to claim 1, wherein In step S1, the sorting of the R files returned from field production is to group the records in the R files returned from field production according to the line numbers and point numbers of the receiving points in the R files returned from field production. When the number of records in the group is ≥2, the record with the highest priority is retained, denoted as OffsetFile, and the records not retained are deleted to obtain the sorted R file; The coordinate replacement is to replace the coordinates of the receiving points in the sorted R file with the actually measured coordinates to obtain the R file after coordinate replacement; The record with the highest priority is the first normal record of this receiving point; Compare the line numbers and point numbers of the receiving points in the R file after coordinate replacement with the line numbers and point numbers of the receiving points in the X file returned from field production. When there are missing receiving point records in the R file after coordinate replacement, after manual investigation, export the correct R file again, import it into the database, and start executing all steps from S1 until there are no missing receiving point records in the R file after coordinate replacement, that is, completing the preliminary quality control of R files.

3. The seismic exploration auxiliary data quality control method according to claim 2, wherein The preliminary quality control of S files is to group the records in the S files returned from field production according to the line numbers and point numbers of the source points in the S files returned from field production, retain the source point record with the largest point index in the group, delete the records not retained, to obtain the retained S file, calculate the coordinate distance of the source points with the same line numbers and point numbers in the retained S file and the S file of the task book. When the distance exceeds the range allowed by the task book, delete this source point from the retained S file, that is, completing the preliminary quality control of S files.

4. The quality control method for seismic exploration auxiliary data according to claim 3, characterized in that When any source point in the S file after preliminary quality control is triggered after the receiving point time in OffsetFile and the line number of the receiving point in OffsetFile needs to be used, the X file returned from field production corresponding to such source points in the S file needs to be modified; The modification of the X file returned from field production corresponding to such excitation points is to extract the line number from the OffsetFile, denoted as Line. According to Line, all records containing this line number are retrieved from the R file after preliminary quality control, denoted as ALLR. The records in ALLR with duplicate line numbers and point numbers as those in the OffsetFile are replaced with the records in the OffsetFile. Then, the point index of ALLR after replacement is incremented by 1 and added to the R file after preliminary quality control. At this time, there are two types of records for the line number Line, one is the record before rectification, denoted as A, and the other is the record after rectification, denoted as B. The receiving point index of the records containing Line in the X file returned from field production is modified to the point index of B to obtain the modified X file. When any excitation point in the S file after preliminary quality control is excited before the receiving point time in the OffsetFile or the line number of the receiving point in the OffsetFile is not required, the X file corresponding to such excitation points does not need to be modified.

5. The seismic exploration auxiliary data quality control method according to claim 4, wherein The X file collation is to extract the corresponding records in the X file according to the line number, point number, and point index of the excitation points in the S file after preliminary quality control, denoted as tempX. The X file refers to the modified X file or the X file returned from field production that does not need to be modified. According to the line number and point number of the excitation points in the S file after preliminary quality control, the corresponding records are extracted from the X file in the task book and stored in the temporary data table together with tempX, denoted as temp. According to the shot point line number, shot point number, receiving line number, starting point number of the receiving line, and ending point number of the receiving line, calculate the number of identical records in temp. When the number of identical records is not 2, conduct manual inspection and import the correct file into the database. Execute all steps starting from S1 until the number of identical records in temp is 2, that is, the preliminary quality control of the X file is completed.

6. The seismic exploration auxiliary data quality control method according to claim 5, wherein The repeated shot quality control is to extract the relationship records from the X files that have completed quality control in the past according to the line number and point number of the S file after preliminary quality control. When the number of extracted relationship file records is 0, it indicates that this shot is not a repeated shot. When the number of extracted relationship file records is not 0 and is consistent with the relationship file record of this shot in tempX, this shot is a repeated shot. Delete this shot point from the S file and X file after preliminary quality control, that is, the repeated shot quality control is completed.

7. The seismic exploration auxiliary data quality control method according to claim 6, characterized in that, The quality control of the excitation point source indicators is to extract the file number from the X file after preliminary quality control according to the line number, point number, and point index of the excitation points extracted from the S file after preliminary quality control. Then, according to the file number, extract the source work indicators from the source vehicle work log file returned from field production. When the source work indicators are unqualified, check whether this excitation point has been excited multiple times. If it has been excited only once, directly delete it from the S file and X file after preliminary quality control. If it is excited more than twice, all records are extracted from the source vehicle work log file according to the line number and point number of the excitation point, the excitation points with qualified source work indicators are selected, the point index of this excitation point is denoted as N, the point index of this excitation point in the S file after preliminary quality control is modified to N, and the point index of the excitation point in the relationship record of this excitation point in the X file after preliminary quality control is also modified to N; If no qualified excitation point can be selected, this excitation point is deleted from the S file after preliminary quality control, that is, the quality control of the source indicators of the excitation point is completed.

8. Use of the seismic exploration auxiliary data quality control method according to claims 1-7 in the warning of the source state, characterized in that After completing the quality control of the source indicators of the excitation point, use LSTM to give early warnings about the source state, including the following steps: P1. Extract all records of the source under study during the study time from the source vehicle work log file after quality control, sort them in ascending order of time, extract the data corresponding to the source state indicators that need to be warned, denoted as Record, perform maximum-minimum normalization on Record by column, the number of columns in Record is denoted as C, the number of rows is denoted as M, and the standardized data is obtained; P2. Divide the standardized data into M training samples, divide the training samples into a training set and a test set, the division ratio of the training set to the test set is (1 - RTT): RTT, the training set contains (1 - RTT) × M training samples, the test set contains RTT × M training samples, and the range of RTT is (0, 1); P3. Apply LSTM to the training samples. At this time, the number of LSTM layers is denoted as Layer, where Layer > 0 and is a positive integer; the number of LSTM neurons in the i-th layer is denoted as j, i is any integer from 1 to Layer, j > 0 and j is a positive integer; the neuron inactivation rate between adjacent layers is denoted as DeadR; Construct a source warning model according to the structure of the input layer, LSTM layers with different numbers of neurons in Layer layers, a Dropout layer inactivated with DeadR, a ReLu activation function connecting the last LSTM layer and the output layer, and the output layer; P4. Use the training set divided in P2 to train the source warning model, use an optimizer to optimize the entire source warning model, use the mean absolute error as the loss function during optimization, input BZ training samples each time during training. When all training samples in the training set have been traversed once, it is one training cycle. When training for Epoch cycles and the value of the loss function < 0.005, the training of this model ends. Subsequently, use the samples in the test set to test the source warning model, calculate the absolute errors of the peak distortion and average distortion between the output value of the source warning model and the expected value of the test set. When the average absolute errors of both are < 5%, stop training; otherwise, train again until the average absolute errors of both are < 5%; The Epoch > 0 and is a positive integer; The BZ is [1, RTT × M] and is a positive integer.

9. Application of the seismic exploration auxiliary data quality control method according to claim 8 in the warning of the source state, characterized in that, The ReLu activation function compares the input value with 0 and takes the maximum value of the two; The optimizer is the Adaptive Moment Estimation optimizer, and the Adaptive Moment Estimation optimizer is the "Adam" optimizer.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program for executing the seismic exploration auxiliary data quality control method according to any one of claims 1 to 9.

Images