Automatic seismic data cleaning method and system with drilling

Abstract

This invention discloses an automatic cleaning method and system for seismic data during excavation. The method involves acquiring seismic data, processing it using a first processing step to determine the optimal reference trace, and then using a second processing step and the optimal reference trace to evaluate the data and obtain a quality score. A data cleaning threshold is determined based on the quality score. First, a preliminary data analysis is conducted to determine the optimal reference trace. Then, all trace data are interferometrically analyzed with the optimal reference trace. The abstract contours of the generated interferometric profiles are extracted, and each trace is analyzed to determine if there is a dominant energy peak. The number of traces with dominant energy peaks is divided by the total number of traces to obtain the quality evaluation result. A high-quality data cleaning threshold is determined based on the seismic data, and this threshold is used to clean high-quality data. Finally, false high-quality data is removed from these selected high-quality data, thus solving the technical problem of inaccurate results in existing seismic data cleaning methods.

Description

Technical Field

[0001] This invention belongs to the field of seismic data cleaning with mining machinery as the seismic source, and relates to data cleaning methods, specifically an automatic method and system for cleaning seismic data during excavation. Background Technology

[0002] When using tunnel boring machines as seismic sources for advanced seismic exploration of coal mine roadways' faces and sides, seismic data is continuously acquired 24 hours a day, resulting in a massive amount of data, of which only a very small percentage is of high quality. Without data cleaning, this inferior data will contaminate the data, wasting computational resources, consuming excessive time, and yielding unsatisfactory results. Therefore, data cleaning is essential. Because seismic data is acquired continuously 24 hours a day, resulting in a huge volume of data, this data cleaning method must be completely free of human intervention. Any cleaning method requiring manual intervention will become overwhelming in the face of a continuous stream of data, causing the vision of real-time processing and real-time imaging to fail.

[0003] Currently, there are two methods for automatic data cleaning. The first method utilizes the strong vibrations throughout the tunnel caused by the belt conveyor during tunneling. By scanning the amplitude of the data track, a threshold is determined to indicate whether the tunneling machine is operating, thus dividing the machine's working and resting periods. However, the tunneling machine does not constantly and forcefully cut the coal face; it also scrapes loose coal from the surface. During these times, the belt conveyor is still running, resulting in larger data track amplitudes, but the corresponding data quality is poor. Therefore, this method cannot guarantee that the selected data will always be of excellent quality. The second method selects one track as a reference track and correlates its signal with the other tracks in the time domain. The data quality is then judged by analyzing the correlation results. While this method has some effectiveness, it has three shortcomings: 1. It is time-domain correlated, and the time consumed increases exponentially with the data length. Longer data blocks will lead to a significant increase in time consumption; 2. It cannot remove highly correlated data caused by large vibration signals other than those caused by tunneling machines, such as micro-seismic events; 3. The entire process requires manual determination of the optimal reference path, manual determination of data cleaning standards, and manual checking for highly correlated data caused by non-tunneling machine vibration sources, which cannot meet the requirements of automatic data processing.

[0004] In summary, existing methods for cleaning seismic data during excavation suffer from inaccurate results, excessive computation time, and the need for significant manual intervention, making it impossible to achieve unattended automated cleaning. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide an automatic cleaning method and system for seismic data during excavation, thereby solving the technical problem of inaccurate results in existing seismic data cleaning methods.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0007] An automatic cleaning method for seismic data during tunneling is based on a seismic data acquisition system, which includes several geophones and signal transmission cables. The geophones are mounted on coal face anchors. The seismic data consists of vibration signals simultaneously collected by multiple geophones when the tunneling machine cuts the coal face, with each geophone collecting a vibration signal called a data channel. The method specifically includes the following steps:

[0008] Step 1: Acquire seismic data, process the seismic data using the first processing method, and determine the optimal reference trace;

[0009] The first process includes a flow and process ;

[0010] The process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment;

[0011] The process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r;

[0012] Step 2: The seismic data is evaluated using the second processing and the optimal reference trace determined in Step 1 to obtain a quality score. The data cleaning threshold is then determined based on the quality score.

[0013] The second process includes the following steps: and process ;

[0014] The process The input consists of multiple two-dimensional raw data segments R and the channel number n of the optimal reference channel determined in step one, and the output consists of multiple second-pulse truncation segments.

[0015] The process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R;

[0016] Step 3: Use the data cleaning threshold obtained in Step 2 to clean the seismic data, obtain high-quality data segments, and then proceed according to the workflow. Remove false high-quality data from high-quality data segments;

[0017] The process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0018] Step one specifically includes the following steps:

[0019] Step 1.1: Lay out the survey lines. After successfully acquiring and uploading the data, select the seismic data from the first day when the tunneling machine is engaged in tunneling activities.

[0020] Step 1.2: The seismic data obtained in Step 1.1 is sliced ​​using a first preset time to obtain multiple two-dimensional raw data segments r. These multiple two-dimensional raw data segments r, along with the trace number n of the trace closest to the tunnel boring machine, are then sequentially sent into the process flow. Output pulse number 1 to extract the segment;

[0021] Step 1.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the first pulse interception section. The pulse segment obtained in step 1.2 is sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input process , thus obtaining the quality scores for all two-dimensional raw data segments r;

[0022] Step 1.4: Select the data segment corresponding to the two-dimensional original data segment r whose quality score is greater than the first threshold;

[0023] Step 1.5: Using the data segment obtained in Step 1.4, each data channel is used as a reference channel to find the best reference channel among all channels, resulting in multiple candidate best reference channels.

[0024] Step 1.6: Calculate the candidate optimal reference track obtained in Step 1.5 using the following formula. The optimal reference track is the candidate track corresponding to its maximum value.

[0025]

[0026]

[0027] in:

[0028] cov(b,j) represents the covariance between the candidate best reference track b and the data track j;

[0029] Indicates the first The variance of the data;

[0030] N represents the total number of data channels;

[0031] This represents the sum of squared correlation coefficients between the candidate best reference channel b and all candidate best reference channels.

[0032] Step two specifically includes the following steps:

[0033] Step 2.1: The seismic data is sliced ​​and segmented using the second preset time to obtain multiple two-dimensional raw data segments R;

[0034] Step 2.2: Sequentially input the multiple two-dimensional raw data segments R obtained in Step 2.1 and the channel number n of the optimal reference channel determined in Step 1 into the process flow. This yields the second pulse segment corresponding to each of the multiple two-dimensional original data segments R;

[0035] Step 2.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the second pulse interception section. The multiple pulse segments obtained in step 2.2 are sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input data flow This yields the quality scores for all two-dimensional raw data segments R, thus obtaining the quality score curve for the seismic data.

[0036] Step 2.4: Determine the data cleaning threshold based on the quality score curve obtained in Step 2.3.

[0037] The specific steps for determining the data cleaning threshold based on the quality scoring curve obtained in step 2.3 are as follows:

[0038] Step 2.4.1: For two-dimensional raw data segments R with a quality score lower than P, they are directly cleaned up; for two-dimensional raw data segments R with a quality score higher than Q, they are determined to be high-quality data segments; for two-dimensional raw data segments R with a quality score in the range of [P, Q], they are determined to be optional data.

[0039] Step 2.4.2: Sort the optional data in the range [P, Q] from largest to smallest according to the quality score, and take the quality score of the top 80% as the data cleaning threshold.

[0040] Step three specifically includes the following steps:

[0041] Step 3.1: Based on the data cleaning threshold obtained in Step 2, remove data segments from the seismic data whose quality scores are lower than the data cleaning threshold to obtain high-quality data segments;

[0042] Step 3.2: Remove false high-quality data from the high-quality data segments obtained in Step 3.1.

[0043] The specific steps for removing false high-quality data from high-quality data segments are as follows:

[0044] Step 3.2.1: Obtain the dominant frequency of microseismic events on the high-quality data segment. The selected high-quality data segments correspond to the two-dimensional raw data segment r and the dominant frequency of the microseismic events. Input process Output the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment;

[0045] Step 3.2.2: Determine whether the quality score obtained in step 3.2.1 exceeds the second threshold. If so, the two-dimensional original data segment r is false high-quality data caused by large vibration signals from non-tunneling machine activities. Set its quality score to -1 and remove it directly.

[0046] process Specifically, the process includes the following steps:

[0047] Q1.1, based on the track number n of the data track closest to the tunneling machine, find the data track corresponding to track number n from the input two-dimensional data segment r, and save it as a reference track. ;

[0048] Q1.2, the data channels in the input two-dimensional data segment r Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ;

[0049] Q2.3, the middle part of the time-domain cross-coherence result obtained in Q1.2 is truncated to a length of... The data yielded a length of The two-dimensional output data, namely the segment of pulse number one;

[0050] process Specifically, the process includes the following steps:

[0051] G1.1, Based on the channel number n of the optimal reference channel determined in step one, find the corresponding data channel from the input two-dimensional data segment R and save it as the reference channel. ;

[0052] G1.2, will transfer the data from the input two-dimensional data segment r to... Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ;

[0053] G1.3, truncates the middle portion of the time-domain coherence result obtained in process 1.2 to a length of... The data yielded a length of The two-dimensional output data, namely the second pulse segment.

[0054] process Specifically, the process includes the following steps:

[0055] Q3.1 Extract contour features from the first pulse segment and abstract them to obtain m abstract contours for each data channel;

[0056] Q3.2, Analyze whether the m data abstract profiles of a data channel obtained in Q3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0057] Q3.3, Perform the operation of Q3.2 on all data channels and calculate the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the first pulse segment.

[0058] The value of the ratio is between [0, 1];

[0059] process Specifically, the process includes the following steps:

[0060] G3.1 extracts contour features from the second pulse segment and abstracts them to obtain m abstract contours for each data channel.

[0061] G 3.2, analyze whether the m data abstract profiles of a data channel obtained from G 3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0062] G 3.3 performs the operation of G 3.2 on all data channels, calculates the ratio of the number of non-zero data channels to the total number of data channels, and this ratio is the quality score of the second pulse segment.

[0063] The value of the ratio is between [0, 1];

[0064] process Specifically, the process includes the following steps:

[0065] W3.1 extracts contour features from the two-dimensional original data segment r corresponding to the high-quality data segment and abstracts them to obtain m data abstract contours for each data channel.

[0066] W3.2, analyze whether the m data abstract profiles of a data channel obtained from W3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0067] W3.3 performs the same operation as W3.2 on all data channels, calculating the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0068] The value of the ratio is between [0, 1].

[0069] The process C specifically includes the following steps:

[0070] Process 1.1, referencing the road Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal;

[0071] Process 1.2, transfer data channels Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal;

[0072] In procedure 1.3, based on the fact that the cross-correlation operation in the time domain is equivalent to the conjugate multiplication of the spectra of the two in the frequency domain, the reference channel is obtained. With Data Frequency domain transformation of time-domain cross-correlation function ;

[0073]

[0074] in:

[0075] It is the frequency domain transformation of the source function;

[0076] It is the complex conjugate of the frequency domain transform of the source function;

[0077] , is the frequency domain conjugate product of the autocorrelation function of the source function, representing the relationship between the source function and the reference trace in the above formula. With Data The effect of cross-correlation function;

[0078] It is from the seismic source S to the detector. The complex conjugate of the Green's function;

[0079] Process 1.4, the reference path is obtained according to the following formula. With Data The expression of the cross-correlation results in the time domain after removing the source influence and normalizing them in the frequency domain ;

[0080]

[0081] in:

[0082] For reference The modulus of the frequency domain transform of the autocorrelation function;

[0083] For data channels The modulus of the frequency domain transform of the autocorrelation function;

[0084] Step 1.5 involves obtaining the results from Step 1.4. Perform inverse Fourier transform to obtain the signal and Cross-correlation results in the time domain after removing the influence of the seismic source and normalizing. ;

[0085] .

[0086] In G3.1, Q3.1, or W3.1, the specific operation to obtain m data abstract contours for each data channel is: using The length is the interval, and the length is... The first pulse segment, the second pulse segment, or the high-quality data segment are cut into m segments;

[0087]

[0088]

[0089] in:

[0090] Indicates the length of time;

[0091] The ceil function rounds up to the nearest integer, meaning it rounds up to the nearest integer if there are no trailing digits. The length portion should be presented as a separate section.

[0092] An automatic cleaning system for seismic data during tunneling, based on the automatic cleaning method for seismic data during tunneling, includes a data acquisition unit, a data quality evaluation unit, and a data cleaning unit connected in sequence.

[0093] The data acquisition unit is used to process the seismic data using a first processing method to determine the optimal reference trace;

[0094] The first process includes a flow and process ;

[0095] The process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment;

[0096] The process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r;

[0097] The data quality evaluation unit is used to evaluate the seismic data using the second processing and the best reference trace, obtain a quality score, and determine the data cleaning threshold based on the quality score.

[0098] The second process includes the following steps: and process ;

[0099] The process The input is multiple two-dimensional raw data segments R and the channel number n of the best reference channel, and the output is multiple second pulse truncation segments;

[0100] The process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R;

[0101] The data cleaning unit is used to clean the seismic data using the obtained data cleaning threshold to obtain high-quality data segments, and based on the process... Remove false high-quality data from high-quality data segments;

[0102] The process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0103] Compared with the prior art, the beneficial technical effects of this invention are:

[0104] (I) In this invention, the optimal reference trace is first determined through data analysis. Then, all trace data are used to perform coherent seismic interferometry with the optimal reference trace. The abstract contour of the generated coherent seismic interferometry profile is extracted. Each trace of the abstract contour is analyzed to determine if there is an absolutely dominant energy peak. The quality evaluation result is obtained by dividing the number of traces with absolutely dominant energy peaks by the total number of traces. Furthermore, a high-quality data cleaning threshold is determined based on the seismic data. This threshold is used to clean high-quality data that can be sent to the next step. Before actually sending it to the next step, false high-quality data caused by large vibration signals such as microseismic events not caused by tunneling machine activities are separately removed from these selected high-quality data. This solves the technical problem of inaccurate results in existing seismic data cleaning methods.

[0105] (II) In this invention, the greater the vibration caused by the tunneling machine cutting the coal face, the stronger its ability to suppress other interferences in the entire arrangement, and the higher the data quality. Therefore, after processing the mutual seismic interference, there are more channels with an absolutely dominant energy peak. Thus, it is very reasonable to use the value of the number of channels with an absolutely dominant energy peak divided by the total number of channels as the quality score (false high-quality data caused by micro-seismic events, etc., are specifically cleaned). On this basis, by automatically determining the best reference channel, automatically determining the data cleaning standard, and automatically cleaning false high-quality data, the entire process greatly reduces the need for manual intervention. It automatically cleans high-quality and true high-quality data from massive seismic data during tunneling, laying the first step foundation for the fully automated processing of seismic data during tunneling. Attached Figure Description

[0106] Figure 1 The automatic data cleaning flowchart of this invention;

[0107] Figure 2 This is the data quality evaluation result of nearly 10 hours of seismic data during tunneling at a coal mine, segmented into 1-minute segments, in this embodiment;

[0108] Figure 3 This is the seismic interferometry result of the seismic data during excavation in this embodiment;

[0109] Figure 4 This is the evaluation result after abstracting and outlining in this embodiment.

[0110] The specific content of the present invention will be further explained in detail below with reference to the embodiments. Detailed Implementation

[0111] It should be noted that, unless otherwise specified, all components in this invention are those known in the art.

[0112] The following are specific embodiments of the present invention. It should be noted that the present invention is not limited to the following specific embodiments. All equivalent modifications made based on the technical solutions of this application fall within the protection scope of the present invention.

[0113] This invention provides an automatic cleaning method for seismic data during tunneling. Based on a seismic data acquisition system, the system includes several geophones and signal transmission cables. The geophones are mounted on coal face anchors. The seismic data consists of vibration signals simultaneously collected by multiple geophones when the tunneling machine cuts the coal face, with each geophone collecting a vibration signal called a data channel. The method specifically includes the following steps:

[0114] Step 1: Acquire seismic data, process the seismic data using the first processing method, and determine the optimal reference trace;

[0115] The first process includes procedures. and process ;

[0116] process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment;

[0117] process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r;

[0118] Step 2: The seismic data is evaluated using the second processing and the optimal reference trace determined in Step 1 to obtain a quality score. The data cleaning threshold is then determined based on the quality score.

[0119] The second process includes procedures. and process ;

[0120] process The input consists of multiple two-dimensional raw data segments R and the channel number n of the optimal reference channel determined in step one, and the output consists of multiple second-pulse truncation segments.

[0121] process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R;

[0122] Step 3: Use the data cleaning threshold obtained in Step 2 to clean the seismic data, obtain high-quality data segments, and then proceed according to the workflow. Remove false high-quality data from high-quality data segments;

[0123] process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0124] In this scheme, the geophone is deployed behind the tunneling machine, approximately 40-50 meters away (this distance is unsuitable for geophone installation). The geophone is mounted on the coal face anchor bolts to receive vibration signals from the coal seam. The larger vibrations that may originate from the coal seam are: 1) micro-vibrations caused by roof rupture; 2) vibrations caused by the tunneling machine cutting through the coal face; and 3) vibrations caused by the belt conveyor transporting coal. Only the vibration caused by the tunneling machine cutting through the coal face is the type of vibration utilized in seismic exploration during tunneling.

[0125] Seismic data is two-dimensional data that is related to time and detector number, while data channels are one-dimensional data that is related to time.

[0126] The tunneling machine has three working states, among which the active coal face cutting state is the seismic data required in this scheme. Therefore, the quality scoring is to select the data segment of the seismic data when the tunneling machine is in the active coal face cutting state.

[0127] (1) Complete shutdown state: At this time, neither the tunneling machine nor the belt conveyor is working, the roadway is relatively quiet, and there are no effective signals.

[0128] (2) Coal cutting state: At this time, the cutting head of the tunneling machine loosens the surface coal and the shovel plate sends the coal to the belt conveyor to the surface. Both the tunneling machine and the belt conveyor are working, but there is only a high-quality signal when the cutting head cuts the hard coal wall. There is no high-quality signal at other times.

[0129] (3) Active coal wall cutting state: At this time, the cutting head of the tunneling machine pushes against the coal wall to actively cut and generate a large number of high-quality signals.

[0130] First, a preliminary data analysis is conducted to determine the optimal reference trace. Then, all trace data are used to perform coherent seismic interferometry with the optimal reference trace. The abstract contours of the generated coherent seismic interferometry profiles are extracted, and each trace within the abstract contour is analyzed to determine if there is a dominant energy peak. The number of traces with dominant energy peaks is divided by the total number of traces to obtain the quality assessment result. Further, a high-quality data cleaning threshold is determined based on the seismic data. This threshold is used to clean high-quality data suitable for the next step. Before actually sending this data to the next step, false high-quality data caused by large vibration signals from microseismic events or other non-tunneling machine activities are removed from these selected high-quality data. This solves the technical problem of inaccurate results in existing seismic data cleaning methods.

[0131] The greater the vibration caused by the tunneling machine cutting through the coal face, the stronger its ability to suppress other interferences across the entire data structure, resulting in higher data quality. Consequently, after processing for mutual seismic interference, there are more channels with an absolutely dominant energy peak. Therefore, using the number of channels with an absolutely dominant energy peak divided by the total number of channels as the quality score is very reasonable (specific cleaning was performed for false high-quality data caused by microseismic events, etc.). Based on this, by automatically determining the optimal reference channel, automatically determining the data cleaning standards, and automatically cleaning false high-quality data, the entire process significantly reduces the need for manual intervention. It automatically cleans high-quality and genuine high-quality data from massive amounts of seismic data collected during tunneling, laying the first step towards fully automated processing of seismic data collected during tunneling.

[0132] Step one specifically includes the following steps:

[0133] Step 1.1: Lay out the survey lines. After successfully acquiring and uploading the data, select the seismic data from the first day when the tunneling machine is engaged in tunneling activities.

[0134] Step 1.2: The seismic data obtained in Step 1.1 is sliced ​​using a first preset time to obtain multiple two-dimensional raw data segments r. These multiple two-dimensional raw data segments r, along with the trace number n of the trace closest to the tunnel boring machine, are then sequentially sent into the process flow. Output pulse number 1 to extract the segment;

[0135] Step 1.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the first pulse interception section. The pulse segment obtained in step 1.2 is sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input process , thus obtaining the quality scores for all two-dimensional raw data segments r;

[0136] Step 1.4: Select the data segment corresponding to the two-dimensional original data segment r whose quality score is greater than the first threshold;

[0137] Step 1.5: Using the data segment obtained in Step 1.4, each data channel is used as a reference channel to find the best reference channel among all channels, resulting in multiple candidate best reference channels.

[0138] Step 1.6: Calculate the candidate optimal reference track obtained in Step 1.5 using the following formula. The optimal reference track is the candidate track corresponding to its maximum value.

[0139]

[0140]

[0141] in:

[0142] cov(b,j) represents the covariance between the candidate best reference track b and the data track j;

[0143] Indicates the first The variance of the data;

[0144] N represents the total number of data channels;

[0145] This represents the sum of squared correlation coefficients between the candidate best reference channel b and all candidate best reference channels.

[0146] In the above technical solution, the first preset time is five minutes, and the first threshold can be selected as 0.5.

[0147] Step two specifically includes the following steps:

[0148] Step 2.1: The seismic data is sliced ​​and segmented using the second preset time to obtain multiple two-dimensional raw data segments R;

[0149] Step 2.2: Sequentially input the multiple two-dimensional raw data segments R obtained in Step 2.1 and the channel number n of the optimal reference channel determined in Step 1 into the process flow. This yields the second pulse segment corresponding to each of the multiple two-dimensional original data segments R;

[0150] Step 2.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the second pulse interception section. The multiple pulse segments obtained in step 2.2 are sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input data flow This yields the quality scores for all two-dimensional raw data segments R, thus obtaining the quality score curve for the seismic data.

[0151] Step 2.4: Determine the data cleaning threshold based on the quality score curve obtained in Step 2.3.

[0152] In the above technical solution, the range of the second preset time is 1 to 10 minutes.

[0153] The data cleaning threshold is determined based on the quality scoring curve obtained in step 2.3 as follows:

[0154] Step 2.4.1: For two-dimensional raw data segments R with a quality score lower than P, they are directly cleaned up; for two-dimensional raw data segments R with a quality score higher than Q, they are determined to be high-quality data segments; for two-dimensional raw data segments R with a quality score in the range of [P, Q], they are determined to be optional data.

[0155] Step 2.4.2: Sort the optional data in the range [P, Q] from largest to smallest according to the quality score, and take the quality score of the top 80% as the data cleaning threshold.

[0156] In the above technical solution, [P, Q] takes the value [0.3, 0.7];

[0157] The working state of the tunneling machine corresponding to the two-dimensional raw data segment R with a quality score lower than P is: completely stopped; the working state of the tunneling machine corresponding to the two-dimensional raw data segment R with a quality score higher than Q is: actively cutting coal wall; the working state of the tunneling machine corresponding to the two-dimensional raw data segment R with a quality score in the range of [P, Q] is: coal cutting.

[0158] Step three specifically includes the following steps:

[0159] Step 3.1: Based on the data cleaning threshold obtained in Step 2, remove data segments from the seismic data whose quality scores are lower than the data cleaning threshold to obtain high-quality data segments;

[0160] Step 3.2: Remove false high-quality data from the high-quality data segments obtained in Step 3.1.

[0161] The specific steps to remove false high-quality data from high-quality data segments are as follows:

[0162] Step 3.2.1: Obtain the dominant frequency of microseismic events on the high-quality data segment. The selected high-quality data segments correspond to the two-dimensional raw data segment r and the dominant frequency of the microseismic events. Input process Output the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment;

[0163] Step 3.2.2: Determine whether the quality score obtained in step 3.2.1 exceeds the second threshold. If so, the two-dimensional original data segment r is false high-quality data caused by large vibration signals from non-tunneling machine activities. Set its quality score to -1 and remove it directly.

[0164] In the above technical solution, the second threshold can be 0.7.

[0165] process Specifically, the process includes the following steps:

[0166] Q1.1, based on the track number n of the data track closest to the tunneling machine, find the data track corresponding to track number n from the input two-dimensional data segment r, and save it as a reference track. ;

[0167] Q1.2, the data channels in the input two-dimensional data segment r Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ;

[0168] Q2.3, the middle part of the time-domain cross-coherence result obtained in Q1.2 is truncated to a length of... The data yielded a length of The two-dimensional output data, namely the segment of pulse number one;

[0169] process Specifically, the process includes the following steps:

[0170] G1.1, Based on the channel number n of the optimal reference channel determined in step one, find the corresponding data channel from the input two-dimensional data segment R and save it as the reference channel. ;

[0171] G1.2, will transfer the data from the input two-dimensional data segment r to... Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ;

[0172] G1.3, truncates the middle portion of the time-domain coherence result obtained in process 1.2 to a length of... The data yielded a length of The two-dimensional output data, namely the second pulse segment.

[0173] process Specifically, the process includes the following steps:

[0174] Q3.1 Extract contour features from the first pulse segment and abstract them to obtain m abstract contours for each data channel;

[0175] Q3.2, Analyze whether the m data abstract profiles of a data channel obtained in Q3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0176] Q3.3, Perform the operation of Q3.2 on all data channels and calculate the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the first pulse segment.

[0177] The value of the ratio is between [0, 1].

[0178] process Specifically, the process includes the following steps:

[0179] G3.1 extracts contour features from the second pulse segment and abstracts them to obtain m abstract contours for each data channel.

[0180] G 3.2, analyze whether the m data abstract profiles of a data channel obtained from G 3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0181] G 3.3, perform the operation of G 3.2 on all data channels, calculate the ratio of the number of non-zero data channels to the total number of data channels, and this ratio is the quality score of the second pulse segment;

[0182] The value of the ratio is between [0, 1].

[0183] process Specifically, the process includes the following steps:

[0184] W3.1 extracts contour features from the two-dimensional original data segment r corresponding to the high-quality data segment and abstracts them to obtain m data abstract contours for each data channel.

[0185] W3.2, analyze whether the m data abstract profiles of a data channel obtained from W3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero.

[0186] W3.3 performs the same operation as W3.2 on all data channels, calculating the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0187] The value of the ratio is between [0, 1].

[0188] In the above technical solutions, in the process and process In the middle, the input pulse segment, It is the direct wave dominant frequency during the coal face cutting phase of the tunneling machine, and its purpose is data quality evaluation; in the process The input is a high-quality data segment. It is the dominant frequency of microseismic events, and its purpose is to eliminate false high-quality data.

[0189] In section 3.2 above, the segment length used for segmentation is... The duration of the strong-energy dominant frequency wave is twice the wavelength of the high-energy dominant frequency wave. Therefore, the high-energy dominant frequency wave portion in the data profile can be summarized using 3-6 abstract sample points. The m abstract data profiles of a data channel obtained in 3.2 are sorted from largest to smallest. The first 6 abstract sample points are used to estimate the amplitude of the strong-energy dominant frequency wave. Beyond 3 times the number of sample points (3*6=18), there is no strong-energy dominant frequency wave amplitude (direct waves or microseismic signals propagate through the surrounding rock without dispersion effects; their wave velocity is relatively stable, meaning their wave train length will not be too long, so this estimation is reasonable). The value ranked 19th is taken as the benchmark. Values ​​less than or equal to twice this value among the m abstract points are set to zero (this threshold is an example and can be slightly increased or decreased). If the entire abstract data channel does not contain data greater than twice this value, it indicates that there is no absolutely dominant energy peak in this channel, meaning there is definitely no direct wave. In this case, the entire abstract data channel is set to zero.

[0190] Process C specifically includes the following steps:

[0191] Process 1.1, referencing the road Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal;

[0192] Process 1.2, transfer data channels Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal;

[0193] In procedure 1.3, based on the fact that the cross-correlation operation in the time domain is equivalent to the conjugate multiplication of the spectra of the two in the frequency domain, the reference channel is obtained. With Data Frequency domain transformation of time-domain cross-correlation function ;

[0194]

[0195] in:

[0196] It is the frequency domain transformation of the source function;

[0197] It is the complex conjugate of the frequency domain transform of the source function;

[0198] , is the frequency domain conjugate product of the autocorrelation function of the source function, representing the relationship between the source function and the reference trace in the above formula. With Data The effect of cross-correlation function;

[0199] It is from the seismic source S to the detector. The complex conjugate of the Green's function;

[0200] Process 1.4, the reference path is obtained according to the following formula. With Data The expression of the cross-correlation results in the time domain after removing the source influence and normalizing them in the frequency domain ;

[0201]

[0202] in:

[0203] For reference The modulus of the frequency domain transform of the autocorrelation function;

[0204] For data channels The modulus of the frequency domain transform of the autocorrelation function;

[0205] Step 1.5 involves obtaining the results from Step 1.4. Perform inverse Fourier transform to obtain the signal and Cross-correlation results in the time domain after removing the influence of the seismic source and normalizing. ;

[0206] .

[0207] In G3.1, Q3.1, or W3.1, the specific operation to obtain m data abstract contours for each data channel is: using The length is the interval, and the length is... The first pulse segment, the second pulse segment, or the high-quality data segment are cut into m segments;

[0208]

[0209]

[0210] in:

[0211] Indicates the length of time;

[0212] The ceil function rounds up to the nearest integer, meaning it rounds up to the nearest integer if there are no trailing digits. The length portion should be presented as a separate section.

[0213] In the above technical solution, the maximum absolute value of the data within the input data segment is used to replace the data in that segment, and the length is... The two-dimensional raw data segment r is abstracted into a two-dimensional data abstract contour of length m.

[0214] An automatic cleaning system for seismic data during tunneling, based on an automatic cleaning method for seismic data during tunneling, includes a data acquisition unit, a data quality evaluation unit, and a data cleaning unit connected in sequence.

[0215] The data acquisition unit is used to process the seismic data using the first processing method to determine the optimal reference trace;

[0216] The first process includes procedures. and process ;

[0217] process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment;

[0218] process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r;

[0219] The data quality evaluation unit is used to evaluate seismic data using the second processing and the best reference trace, obtain a quality score, and determine the data cleaning threshold based on the quality score.

[0220] The second process includes procedures. and process ;

[0221] process The input is multiple two-dimensional raw data segments R and the channel number n of the best reference channel, and the output is multiple second pulse truncation segments;

[0222] process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R;

[0223] The data cleaning unit is used to clean the seismic data using the obtained data cleaning threshold, obtain high-quality data segments, and then process them according to the workflow. Remove false high-quality data from high-quality data segments;

[0224] process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

[0225] Example:

[0226] For nearly 10 hours of seismic data from a coal mine, data quality was evaluated in 1-minute segments. The evaluation results are shown below. Figure 2 The evaluation results clearly distinguish between the periods when the tunneling machine actively cuts the coal face and the periods when it stops.

[0227] After taking a segment of seismic data during excavation, obtaining the interferometric profile of the coherent seismic data, extracting the contour features, and then analyzing each trace to determine if there is an absolutely dominant energy peak, the results are as follows: Figures 3-4 . Figure 3 It is the result of cross-interference processing of the seismic data during excavation. Figure 4 yes Figure 3 After abstracting and outlining, and then evaluating, weak energy clusters are set to zero, while the dominant energy clusters are retained. As shown in the figure, this invention is robust, effectively identifying channels without relevant energy peaks, and providing a sufficiently reliable evaluation of the data block quality.

Claims

1. An automatic cleaning method for seismic data during tunneling, based on a seismic data acquisition system, wherein the seismic data acquisition system includes several geophones and signal transmission cables, the geophones being installed on coal face anchors; the seismic data consists of vibration signals simultaneously acquired by multiple geophones when the tunneling machine cuts the coal face, with the vibration signal acquired by one geophone referred to as one data channel, characterized in that... Specifically, the following steps are included: Step 1: Acquire seismic data, process the seismic data using the first processing method, and determine the optimal reference trace; The first process includes a flow and process ; The process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment; The two-dimensional raw data segment r is obtained by slicing the seismic data using a first preset time. The process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r; Step 2: The seismic data is evaluated using the second processing and the optimal reference trace determined in Step 1 to obtain a quality score. The data cleaning threshold is then determined based on the quality score. The second process includes the following steps: and process ; The process The input consists of multiple two-dimensional raw data segments R and the channel number n of the optimal reference channel determined in step one, and the output consists of multiple second-pulse truncation segments. The two-dimensional raw data segment R is obtained by slicing the seismic data using a second preset time. The process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R; Step 3: Use the data cleaning threshold obtained in Step 2 to clean the seismic data, obtain high-quality data segments, and then proceed according to the process. Remove false high-quality data from high-quality data segments; The process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

2. The automatic cleaning method for seismic data during excavation as described in claim 1, characterized in that, Step one specifically includes the following steps: Step 1.1: Lay out the survey lines. After successfully acquiring and uploading the data, select the seismic data from the first day when the tunneling machine is engaged in tunneling activities. Step 1.2: The seismic data obtained in Step 1.1 is sliced ​​using a first preset time to obtain multiple two-dimensional raw data segments r. These multiple two-dimensional raw data segments r, along with the trace number n of the trace closest to the tunnel boring machine, are then sequentially sent into the process flow. Output pulse number 1 to extract the segment; Step 1.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the first pulse interception section. The pulse segment obtained in step 1.2 is sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input process , thus obtaining the quality scores for all two-dimensional raw data segments r; Step 1.4: Select the data segment corresponding to the two-dimensional original data segment r whose quality score is greater than the first threshold; Step 1.5: Using the data segment obtained in Step 1.4, each data channel is used as a reference channel to find the best reference channel among all channels, resulting in multiple candidate best reference channels. Step 1.6: Calculate the candidate optimal reference track obtained in Step 1.5 using the following formula. The optimal reference track is the candidate track corresponding to its maximum value. in: cov(b,j) represents the covariance between the candidate best reference track b and the data track j; Indicates the first The variance of the data; N represents the total number of data channels; This represents the sum of squared correlation coefficients between the candidate best reference channel b and all candidate best reference channels.

3. The automatic cleaning method for seismic data during excavation as described in claim 1, characterized in that, Step two specifically includes the following steps: Step 2.1: The seismic data is sliced ​​and segmented using the second preset time to obtain multiple two-dimensional raw data segments R; Step 2.2: Sequentially input the multiple two-dimensional raw data segments R obtained in Step 2.1 and the channel number n of the optimal reference channel determined in Step 1 into the process flow. This yields the second pulse segment corresponding to each of the multiple two-dimensional original data segments R; Step 2.3: Obtain the main frequency of the direct wave when the tunneling machine cuts the coal face on the second pulse interception section. The multiple pulse segments obtained in step 2.2 are sequentially compared with the main frequency of the direct wave when the tunneling machine cuts the coal face. Input data flow This yields the quality scores for all two-dimensional raw data segments R, thus obtaining the quality score curve for the seismic data. Step 2.4: Determine the data cleaning threshold based on the quality score curve obtained in Step 2.

3.

4. The automatic cleaning method for seismic data during excavation as described in claim 3, characterized in that, The specific steps for determining the data cleaning threshold based on the quality scoring curve obtained in step 2.3 are as follows: Step 2.4.1: For two-dimensional raw data segments R with a quality score lower than P, they are directly cleaned up; for two-dimensional raw data segments R with a quality score higher than Q, they are determined to be high-quality data segments; for two-dimensional raw data segments R with a quality score in the range of [P, Q], they are determined to be optional data. Step 2.4.2: Sort the optional data in the range [P, Q] from largest to smallest according to the quality score, and take the quality score of the top 80% as the data cleaning threshold.

5. The automatic cleaning method for seismic data during excavation as described in claim 1, characterized in that, Step three specifically includes the following steps: Step 3.1: Based on the data cleaning threshold obtained in Step 2, remove data segments from the seismic data whose quality scores are lower than the data cleaning threshold to obtain high-quality data segments; Step 3.2: Remove false high-quality data from the high-quality data segments obtained in Step 3.

1.

6. The automatic cleaning method for seismic data during excavation as described in claim 5, characterized in that, The specific steps for removing false high-quality data from high-quality data segments are as follows: Step 3.2.1: Obtain the dominant frequency of microseismic events on the high-quality data segment. The selected high-quality data segments correspond to the two-dimensional raw data segment r and the dominant frequency of the microseismic events. Input process Output the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment; Step 3.2.2: Determine whether the quality score obtained in step 3.2.1 exceeds the second threshold. If so, the two-dimensional original data segment r is false high-quality data caused by large vibration signals from non-tunneling machine activities. Set its quality score to -1 and remove it directly.

7. The automatic cleaning method for seismic data during excavation as described in claim 1, characterized in that, process Specifically, the process includes the following steps: Q1.1, based on the track number n of the data track closest to the tunneling machine, find the data track corresponding to track number n from the input two-dimensional data segment r, and save it as a reference track. ; Q1.2, the data channels in the input two-dimensional data segment r Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ; Q2.3, the middle part of the time-domain cross-coherence result obtained in Q1.2 is truncated to a length of... The data yielded a length of The two-dimensional output data, namely the segment of pulse number one; process Specifically, the process includes the following steps: G1.1, Based on the channel number n of the optimal reference channel determined in step one, find the corresponding data channel from the input two-dimensional data segment R and save it as the reference channel. ; G1.2, will transfer the data from the input two-dimensional data segment r to... Respectively with reference Perform the operation in process C to obtain the reference path. With Data The normalized time-domain mutual coherence results ; G1.3, truncates the middle portion of the time-domain coherence result obtained in process 1.2 to a length of... The data yielded a length of The two-dimensional output data, namely the segment of pulse number two; The process C specifically includes the following steps: Process 1.1, referencing the road Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal; Process 1.2, transfer data channels Transform to the frequency domain, denoted as ,in It is from the seismic source S to the detector. Green's function, It is a frequency domain transformation of the seismic source signal; In procedure 1.3, based on the fact that the cross-correlation operation in the time domain is equivalent to the conjugate multiplication of the spectra of the two in the frequency domain, the reference channel is obtained. With Data Frequency domain transformation of time-domain cross-correlation function ; in: It is the frequency domain transformation of the source function; It is the complex conjugate of the frequency domain transform of the source function; , is the frequency domain conjugate product of the autocorrelation function of the source function, representing the relationship between the source function and the reference trace in the above formula. With Data The effect of cross-correlation function; It is from the seismic source S to the detector. The complex conjugate of the Green's function; Process 1.4, the reference path is obtained according to the following formula. With Data The expression of the cross-correlation results in the time domain after removing the source influence and normalizing them in the frequency domain ; in: For reference The modulus of the frequency domain transform of the autocorrelation function; For data channels The modulus of the frequency domain transform of the autocorrelation function; Step 1.5 involves obtaining the results from Step 1.

4. Perform inverse Fourier transform to obtain the signal and Cross-correlation results in the time domain after removing the influence of the seismic source and normalizing. ; 。 8. The automatic cleaning method for seismic data during excavation as described in claim 1, characterized in that, process Specifically, the process includes the following steps: Q3.1 Extract contour features from the first pulse segment and abstract them to obtain m abstract contours for each data channel; Q3.2, Analyze whether the m data abstract profiles of a data channel obtained in Q3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero. Q3.3, Perform the operation of Q3.2 on all data channels and calculate the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the first pulse segment. The value of the ratio is between [0, 1]; process Specifically, the process includes the following steps: G3.1 extracts contour features from the second pulse segment and abstracts them to obtain m abstract contours for each data channel. G 3.2, analyze whether the m data abstract profiles of a data channel obtained from G 3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero. G 3.3 performs the operation of G 3.2 on all data channels, calculates the ratio of the number of non-zero data channels to the total number of data channels, and this ratio is the quality score of the second pulse segment. The value of the ratio is between [0, 1]; process Specifically, the process includes the following steps: W3.1 extracts contour features from the two-dimensional original data segment r corresponding to the high-quality data segment and abstracts them to obtain m data abstract contours for each data channel. W3.2, analyze whether the m data abstract profiles of a data channel obtained from W3.1 have an absolutely dominant energy peak. If there is an absolutely dominant energy peak, then the data abstract profile only retains the energy peak. If there is no absolutely dominant energy peak, then this data channel is set to zero. W3.3 performs the same operation as W3.2 on all data channels, calculating the ratio of the number of non-zero data channels to the total number of data channels. This ratio is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment. The value of the ratio is between [0, 1].

9. The automatic cleaning method for seismic data during excavation as described in claim 8, characterized in that, In G3.1, Q3.1, or W3.1, the specific operation to obtain m data abstract contours for each data channel is: using The length is the interval, and the length is... The first pulse segment, the second pulse segment, or the high-quality data segment are cut into m segments; in: Indicates the length of time; The ceil function rounds up to the nearest integer, meaning it rounds up to the nearest integer if there are no trailing digits. The length portion should be presented as a separate section.

10. An automatic cleaning system for seismic data during tunneling, based on the automatic cleaning method for seismic data during tunneling according to any one of claims 1 to 9, comprising a data acquisition unit, a data quality evaluation unit, and a data cleaning unit connected in sequence; The data acquisition unit is used to process the seismic data using a first processing method to determine the optimal reference trace; The first process includes a flow and process ; The process The input is a two-dimensional raw data segment r and the channel number n of the data channel closest to the tunneling machine, and the output is the first pulse segment; The two-dimensional raw data segment r is obtained by slicing the seismic data using a first preset time. The process The input consists of the first pulse segment and the main frequency of the direct wave when the tunneling machine cuts the coal face. The output is the quality score of the original two-dimensional data segment r; The data quality evaluation unit is used to evaluate the seismic data using the second processing and the best reference trace, obtain a quality score, and determine the data cleaning threshold based on the quality score. The second process includes the following steps: and process ; The process The input is multiple two-dimensional raw data segments R and the channel number n of the best reference channel, and the output is multiple second pulse truncation segments; The two-dimensional raw data segment R is obtained by slicing the seismic data using a second preset time. The process The input consists of multiple second-pulse truncation segments and the direct wave main frequency during the tunneling machine's cutting of the coal face. The output is the quality score of the original two-dimensional data segment R; The data cleaning unit is used to clean the seismic data using the obtained data cleaning threshold to obtain high-quality data segments, and based on the process... Remove false high-quality data from high-quality data segments; The process The input consists of the two-dimensional raw data segment r corresponding to the high-quality data segment and the dominant frequency of the microseismic event. The output is the quality score of the two-dimensional original data segment r corresponding to the high-quality data segment.

Images