Small fault identification method and device

Abstract

The invention discloses a minor fault identification method and device. The minor fault identification method comprises the following steps of: analyzing seismic attributes of acquired seismic data of a research area by adopting an optimal analysis method to obtain corresponding to-be-predicted fault feature information; and inputting the to-be-predicted fault feature information into the minor fault identification model to obtain a minor fault identification result. By constructing the minor fault identification model, the minor fault can be automatically identified, the minor fault identification precision is high, the efficiency is high, and the success rate of geological exploration can be improved.

Description

Technical Field

[0001] The present invention relates to the field of geological exploration, and in particular to a method and device for identifying small faults. Background Art

[0002] Fault identification is a critical issue in oilfield exploration and development. Traditional fault identification methods rely primarily on manual interpretation, resulting in low efficiency and accuracy. In recent years, with the development of artificial intelligence (AI), a growing number of researchers have begun exploring its application for fault identification. However, current methods often only consider a single seismic data set and fail to fully exploit the quantitative attribute information from multiple seismic data sets, resulting in low accuracy and efficiency. Therefore, developing a method and device for identifying small faults is of great practical significance and value. Summary of the Invention

[0003] The present invention provides a method and device for identifying small faults to replace traditional fault identification methods. Based on artificial intelligence, the method improves the efficiency and accuracy of small fault identification, so as to better carry out detailed description of oil reservoirs.

[0004] In a first aspect, the present invention provides a method for identifying a small fault, comprising the following steps:

[0005] The seismic data of the study area are analyzed using the optimal analysis method to analyze the seismic attributes and obtain the corresponding fault characteristic information to be predicted;

[0006] Input the characteristic information of the fault to be predicted into the small fault identification model to obtain a small fault identification result; wherein the small fault identification model is obtained by the following method:

[0007] Based on the seismic data, various preset analysis methods are used to analyze seismic attributes to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes of coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume;

[0008] According to the sample data corresponding to each of the preset analysis methods, a pre-established deep learning model is trained to obtain a small fault identification training result;

[0009] Evaluate the small fault identification training result by presetting multiple evaluation parameters to obtain corresponding evaluation results;

[0010] According to the evaluation results corresponding to each of the evaluation parameters, the preferred analysis method is screened from the preset analysis methods;

[0011] Retraining the optimal deep learning model corresponding to the optimal analysis method and performing model optimization;

[0012] The training and optimization process of the preferred deep learning model is repeated until the model meets the preset evaluation conditions, thereby obtaining the small fault identification model.

[0013] In one or some optional implementations of the embodiment of the present application, after inputting the characteristic information of the fault to be predicted into the small fault identification model and obtaining the small fault identification result, the method further includes:

[0014] The small fault identification results are explained using AI visualization technology and spatially combined and displayed.

[0015] One or some optional implementations of the embodiments of the present application may further include:

[0016] The seismic data is obtained by:

[0017] Seismic wave data collected by a seismograph in a study area is obtained, and the seismic wave data is processed to obtain the seismic data.

[0018] One or some optional implementations of the embodiments of the present application may further include:

[0019] The seismic data is preprocessed by constructing a guided filtering method until the seismic data meets the resolution requirement.

[0020] In one or some optional implementations of the embodiments of the present application, the evaluation parameters include accuracy, recall, precision, F1 score and AUC-ROC;

[0021] The evaluating the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result includes:

[0022] The precision, recall rate, accuracy, F1 score and AUC-ROC are used to evaluate the small fault recognition training results to obtain corresponding evaluation results.

[0023] In a second aspect, the present invention provides a method for training a small fault recognition model, comprising the following steps:

[0024] Based on the seismic data, various preset analysis methods are used to analyze seismic attributes to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes of coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume;

[0025] According to the sample data corresponding to each of the preset analysis methods, a pre-established deep learning model is trained to obtain a small fault identification training result;

[0026] Evaluate the small fault identification training result by presetting multiple evaluation parameters to obtain corresponding evaluation results;

[0027] According to the evaluation results corresponding to each of the evaluation parameters, a preferred analysis method is screened from the preset analysis methods;

[0028] Retraining the optimal deep learning model corresponding to the optimal analysis method and performing model optimization;

[0029] The training and optimization process of the preferred deep learning model is repeated until the model meets the preset evaluation conditions, thereby obtaining the small fault identification model.

[0030] In a third aspect, the present invention provides a device for identifying small faults, comprising:

[0031] An analysis module is configured to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume;

[0032] A first training module is used to train a pre-established deep learning model based on sample data corresponding to each of the preset analysis methods to obtain a small fault identification training result;

[0033] A first evaluation module is used to evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result;

[0034] A optimization module, configured to select the preferred analysis method from the preset analysis methods according to the evaluation results corresponding to each evaluation parameter;

[0035] A second training module is used to retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization;

[0036] a second evaluation module, configured to evaluate the preferred deep learning model; if the preferred deep learning model does not meet the preset evaluation conditions, causing the second training module to re-execute the model training process; if the preferred deep learning model meets the preset evaluation conditions, obtaining a small fault recognition model;

[0037] An acquisition module is used to analyze seismic attributes of the acquired seismic data of the study area using an optimal analysis method to obtain corresponding fault feature information to be predicted;

[0038] The identification module is used to input the characteristic information of the fault to be predicted into the small fault identification model to obtain a small fault identification result.

[0039] In a fourth aspect, the present invention provides a training device for a small fault recognition model, comprising:

[0040] An analysis module is configured to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume;

[0041] A first training module is used to train a pre-established deep learning model based on sample data corresponding to each of the preset analysis methods to obtain a small fault identification training result;

[0042] A first evaluation module is used to evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result;

[0043] A optimization module, configured to select the preferred analysis method from the preset analysis methods according to the evaluation results corresponding to each evaluation parameter;

[0044] A second training module is used to retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization;

[0045] The second evaluation module is used to evaluate the preferred deep learning model. If the preferred deep learning model does not meet the preset evaluation conditions, the second training module is enabled to re-execute the model training process. If the preferred deep learning model meets the preset evaluation conditions, a small fault recognition model is obtained.

[0046] In a fifth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the small fault identification method described in the first aspect, and / or the small fault identification model training method described in the second aspect.

[0047] In a sixth aspect, the present invention provides an electronic device comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0048] Memory for storing computer programs;

[0049] The processor is configured to implement the small fault identification method described in the first aspect and / or the small fault identification model training method described in the second aspect when executing the program stored in the memory.

[0050] The present invention provides a method and apparatus for identifying small faults. By analyzing seismic attributes using a preferred analysis method for seismic data from a study area, corresponding characteristic information of the fault to be predicted is obtained. The characteristic information of the fault to be predicted is then input into a small fault identification model, thereby automatically identifying small faults. When constructing the small fault identification model, the seismic attributes are analyzed using various preset analysis methods to obtain characteristic information of the corresponding small fault as sample data. Each of the preset analysis methods includes an analysis method combining at least two attributes: a coherence volume, an ant tracking volume, a small window coherence volume, a diffusion filter, and a frequency-divided ant volume. This allows the sample data to contain at least two seismic attributes. This solves the problem of difficulty and low accuracy in identifying small faults in the prior art, which is caused by the analysis of a single seismic attribute. Small fault identification is highly accurate and efficient. Furthermore, by comprehensively analyzing multiple seismic attributes and selecting an analysis method that is suitable for small fault identification, the structure, lithology, and fluid distribution of the stratum can be more accurately revealed, providing an important basis for identifying small faults. Small fault identification is highly accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the prior art description. The drawings mentioned here are used to provide a further understanding of the present invention, constitute part of this application, and do not constitute a limitation of the present invention. In the drawings:

[0052] Figure 1 A schematic flow chart of a method for identifying small faults provided by one embodiment of the present invention;

[0053] Figure 2 A schematic flow chart of a method for identifying small faults provided by one embodiment of the present invention;

[0054] Figure 3 An example diagram of quantitative seismic attributes provided by an embodiment of the present invention;

[0055] Figure 4 An example diagram of a coherence volume provided by an embodiment of the present invention;

[0056] Figure 5 This is an example diagram of ant tracking provided by an embodiment of the present invention;

[0057] Figure 6 This is an example diagram of a coherence volume of an hour window provided by an embodiment of the present invention;

[0058] Figure 7 This is an example diagram of diffusion filtering provided by an embodiment of the present invention;

[0059] Figure 8 This is an example diagram of a frequency-divided ant body provided by an embodiment of the present invention;

[0060] Figure 9 This is an example diagram of a multi-attribute combination provided by an embodiment of the present invention;

[0061] Figure 10 An example diagram of five evaluation parameters provided in one embodiment of the present invention;

[0062] Figure 11 A schematic diagram of a training device for a small fault recognition model provided by one embodiment of the present invention;

[0063] Figure 12 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The following describes the embodiments of the present invention in detail with reference to the accompanying drawings and examples, so that the present invention can fully understand how to apply technical means to solve technical problems and achieve technical effects, and thus implement the invention accordingly. It should be noted that, as long as no conflict exists, the various embodiments of the present invention and the various features therein can be combined with each other, and the resulting technical solutions are all within the scope of protection of the present invention.

[0065] Additionally, the steps shown in the flowcharts of the accompanying drawings may be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in an order different from that shown.

[0066] A fault is a displacement phenomenon caused by the fracture of a rock layer or rock mass in the earth's crust when, under stress, the rock's ultimate strength falls below the pressure. Fault diagnosis plays a crucial role in the safety of mineral production. Conventional fault identification is achieved by observing characteristics such as amplitude, phase, and time difference on seismic profiles. However, small faults exhibit minimal changes in the time profile, making them difficult to observe with the naked eye. Consequently, fault identification results are subject to subjective factors. Therefore, existing techniques typically use seismic attributes to identify faults.

[0067] Seismic attributes are kinematic, dynamic, geometric, and statistical characteristics obtained through mathematical calculations of seismic data. For faults with different fault throws, the response effects of seismic attributes are also different. Taking the variance attribute as an example, as the fault throw increases, the corresponding variance value also increases. The variance at a fault with a fault throw of 1 meter is on the order of 10 -6 The variance of the fault at a fault distance of 20 meters is on the order of 10 -2, it can be seen that the variance magnitudes of the two methods differ significantly. Variance attributes can reveal the presence of faults, but due to the significant difference in variance magnitude between large and small faults, large faults can easily overshadow small fault information. In reality, the variation in fault throws is very complex, often accompanied by variations in various spacings. This can lead to a situation where large values ​​are emphasized and small values ​​are ignored, resulting in unclear identification of small faults and reducing the overall effectiveness of fault identification. Therefore, the accuracy of small fault identification is low using a single seismic attribute analysis.

[0068] Based on this, in order to solve the above problems, the inventors proposed a small fault identification method and device. The small fault identification model established based on seismic attribute analysis of multi-attribute combination can automatically identify small faults and improve the identification accuracy of small faults.

[0069] Example 1

[0070] refer to Figure 1 This embodiment provides a method for identifying a small fault, comprising the following steps:

[0071] Step S1: Analyze the seismic attributes of the acquired seismic data of the study area using a preferred analysis method to obtain corresponding fault feature information to be predicted.

[0072] Step S2: Input the characteristic information of the fault to be predicted into the small fault identification model to obtain the small fault identification result.

[0073] The preferred analysis method in step S1 is determined during the training of the small fault identification model in step S2. The preferred analysis method can be selected from analysis methods combining at least two attributes including coherence volume, ant tracking, small window coherence volume, diffusion filtering, and frequency division ant volume. Figure 2 The small fault recognition model in step 2 is trained in the following way:

[0074] Step S201: Based on the seismic data, various preset analysis methods are used to analyze seismic attributes and obtain characteristic information of corresponding small faults as sample data. Each preset analysis method includes an analysis method that combines at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume.

[0075] In the embodiment of the present invention, the seismic data in step S201 above can be obtained in the following manner:

[0076] Seismic wave data is collected in the study area using a seismograph, and the seismic wave data is processed to obtain seismic data, which contains information such as seismic waveform, amplitude, frequency, and phase. The seismic data is quality controlled to determine whether it meets the standards of "three highs and one accuracy" (i.e., high signal-to-noise ratio, high resolution, high fidelity, and accurate imaging). In this process, the structure-guided filtering method can be used to preprocess the seismic data until the seismic data meets the resolution requirements, that is, it can meet the resolution requirements for extracting seismic attributes using various preset analysis methods. The method of preprocessing seismic data can adopt conventional methods, such as denoising. The specific implementation process can refer to the detailed description of the existing technology and is not described in detail here.

[0077] Seismic attribute analysis is a technique used to interpret and analyze seismic data. It quantifies certain characteristics within seismic data to help geologists and geophysicists better understand the structure, lithology, and fluid distribution of strata. The principle of seismic attribute analysis is primarily based on the physical process by which seismic waves propagate through underground strata. As seismic waves propagate through underground strata, they are affected by the physical properties of the strata (such as density and elastic modulus), causing changes in the amplitude, phase, and frequency of the waves. Attribute analysis of seismic data can extract this information, revealing the physical properties and structural characteristics of the strata.

[0078] See Figure 3 , earthquake attributes can be divided into many categories, mainly including the following categories:

[0079] 1) Amplitude attributes: reflect the changes in seismic wave amplitude, such as amplitude, energy, envelope, etc. Amplitude attributes can be used to identify formation characteristics such as lithology and fluid saturation. Figure 3 The root mean square amplitude attribute is a type of amplitude attribute. The root mean square amplitude is positively correlated with the rock density. Therefore, the rock density can be judged according to the size of the root mean square amplitude, and then the lithology and lithofacies can be distinguished.

[0080] 2) Time attributes: reflect changes in seismic wave propagation time, such as round-trip time, delay time, etc. Time attributes can be used to identify velocity structure and thickness changes of strata.

[0081] 3) Frequency attributes: reflect the changes in seismic wave frequency, such as spectrum, bandwidth, etc. Frequency attributes can be used to identify the detailed structure of the stratum and the physical properties of the rock. Figure 2 The instantaneous frequency in the figure represents the frequency value of the seismic signal at each time point. The instantaneous frequency can reflect the lithological characteristics, sedimentary characteristics, and pore fluid properties, and can also be used to identify fault areas.

[0082] 4) Phase attributes: reflect changes in the phase of seismic waves, such as phase angle, instantaneous phase, etc. Phase attributes can be used to identify interfaces and scattering characteristics of strata.

[0083] 5) Waveform attributes: reflect changes in seismic waveforms, such as waveform similarity and waveform classification. Waveform attributes can be used to identify changes in the structure and lithology of strata.

[0084] Among them, median filtering is a nonlinear signal processing technology based on sorting statistical theory that can effectively suppress noise. It can completely eliminate spike pulses and is used to eliminate high-frequency noise in reflected seismic data. It also completely eliminates step and harmonic waves, which is beneficial to improving the lateral resolution of seismic data. The above seismic attributes are extracted from the filtered seismic data, such as Figure 2 The median filter attribute in is used to identify small faults.

[0085] The 3D edge enhancement attribute can highlight the edges of small faults by enhancing the edges and contours in the image, making them easier to detect and identify, thereby improving the visibility and identifiability of small faults. By combining with other seismic attributes, small faults can be identified more accurately, providing more accurate geological information for oil and gas exploration and development. Figure 3 The 3D curvature body properties in can be used to describe small folds and deflections, which is beneficial for the characterization of small faults. Figure 3 The orientation attributes in the stratum can provide information about the lithology and structure of the stratum, including the strike, dip and inclination of the stratum. This information can be used to identify the existence and location of small faults.

[0086] Therefore, through comprehensive analysis of seismic attributes, the problem of poor accuracy in identifying small faults caused by single seismic attribute analysis in existing technologies can be solved, and the structure, lithology and fluid distribution of the strata can be revealed more accurately, providing an important basis for the identification of small faults.

[0087] The analysis methods for earthquake attributes include the analysis methods of at least two attribute combinations including coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume. Figure 4 , is a schematic diagram of the coherence volume. Coherence technology can be used to analyze the correlation of seismic signals and, combined with seismic attribute analysis, identify the boundaries and locations of faults. Figure 5 , a schematic diagram of ant tracking technology, is a complex earthquake attribute algorithm based on swarm intelligence. By simulating the foraging behavior of ants, it can automatically identify the boundaries and shapes of small faults in seismic data. Figure 6 , is a schematic diagram of the hour window coherence technology, which can analyze the correlation of seismic signals within a small time window and identify the boundaries and locations of small faults by combining seismic attribute analysis. Figure 7, is a schematic diagram of the diffusion filtering technology, which can filter out high-frequency noise in seismic signals while retaining useful signals. Based on the seismic data obtained from the processed seismic signals, further seismic attribute analysis is performed to enhance the recognition accuracy of small faults. Figure 8 , is a schematic diagram of the frequency division ant volume technology, which can divide the seismic signal into different frequencies and then extract the seismic data in each frequency band for ant volume attribute extraction. Ant volume attribute is a seismic attribute that uses ant tracking technology to identify the fault boundary and shape in different frequency bands. Figure 9 Figure 2 is a schematic diagram of a multi-attribute combination, specifically a seismic attribute analysis method combining coherence volume and ant tracking. Seismic attribute techniques such as coherence volume and ant tracking are highly indicative of faults. However, in some extremely complex fault-block areas with fragmented strata, identifying small faults based solely on a single seismic attribute is extremely difficult and suffers from low accuracy. This multi-attribute analysis method, by comprehensively analyzing multiple seismic attributes, can address these issues and accurately identify small faults. Therefore, based on seismic attribute adaptability analysis, a method for identifying small faults using a combination of at least two seismic attributes can be established to improve the accuracy of small fault identification.

[0088] Step S202: Based on the sample data corresponding to each preset analysis method, a pre-established deep learning model is trained to obtain a small fault identification training result. The deep learning model may be a convolutional neural network, a recurrent neural network, or other applicable deep learning models.

[0089] Step S203: Evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result.

[0090] In step S203, specifically, refer to Figure 10 ,Preferably use precision, recall, precision, F1 score and AUC-ROC to evaluate the small fault recognition training results ,to obtain the corresponding evaluation results.

[0091] Precision refers to the ratio of the number of samples correctly predicted by the model to the total number of samples. In a binary classification problem, precision is equal to the number of true positives (TP) and true negatives (TN) divided by the number of all samples. Precision is a basic indicator for measuring the overall predictive ability of the model.

[0092] Recall, also known as sensitivity or true positive rate, is the ratio of the number of positive samples correctly identified by the model to the number of all actual positive samples. It is an important indicator to measure the model's ability to identify positive samples. In binary classification problems, recall is equal to the number of true positives (TP) divided by the number of all actual positive samples (i.e., TP + FN).

[0093] Precision refers to the ratio of the number of positive samples correctly identified by the model to the total number of samples identified as positive samples by the model. It is an important indicator to measure the accuracy of the model in predicting positive samples. In the binary classification problem, precision is equal to the number of true positives (TP) divided by the number of all samples predicted as positive by the model (i.e. TP + FP).

[0094] The F1 score refers to the harmonic mean of precision and recall, which takes both precision and recall into account. It is an important indicator for measuring the overall performance of a model. The F1 score ranges from 0 to 1. A larger value indicates better model performance.

[0095] AUC-ROC refers to the area under the receiver operating characteristic curve (AUC). The ROC curve is plotted with the false positive rate (FPR) on the horizontal axis and the true positive rate (TPR) on the vertical axis. AUC values ​​range from 0 to 1, with larger values ​​indicating better model performance. AUC-ROC is a comprehensive indicator that measures model performance at different thresholds.

[0096] Step S204: Filter and obtain a preferred analysis method from various preset analysis methods according to the evaluation results corresponding to each evaluation parameter.

[0097] Step S205: Retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization.

[0098] In step S205, by analyzing the decision-making process of the preferred deep learning model and understanding how the model extracts feature recognition parameters and identifies minor faults based on them, the user is helped to understand the advantages and disadvantages of the preferred deep learning model. Based on these disadvantages, the preferred deep learning model can be further optimized. Specifically, model optimization can be performed using backpropagation and optimization algorithms, such as gradient descent, to improve the accuracy of minor fault identification.

[0099] Step S206: Repeat the training and optimization process of the optimal deep learning model until the model meets the preset evaluation conditions, thereby obtaining a small fault recognition model.

[0100] In step S206, after the optimization process, the selected deep learning model is evaluated. The model's prediction performance is determined by evaluating its prediction results. Evaluation parameters include accuracy, recall, precision, F1 score, and AUC-ROC. Users can set preset thresholds for each evaluation parameter. This helps us understand the model's performance on the small fault identification task and further optimize the prediction model parameters. When the evaluation results for the evaluation parameters reach the preset thresholds, the model meets the preset evaluation criteria, and the training and optimization process for the selected deep learning model is terminated, resulting in a small fault identification model.

[0101] In step S2, the small fault identification model obtains small fault identification results with high accuracy. After obtaining the small fault identification results, AI visualization technology can be used to interpret the small fault identification results and spatially combine and display them. First, data visualization can be achieved. Using data visualization technology, the identified small faults can be displayed in the form of graphics or images, helping users quickly understand the spatial distribution characteristics of the faults.

[0102] In an embodiment of the present invention, in the analysis of small fault results, by analyzing the decision-making process of the model, we can understand how the model extracts feature recognition parameters from multi-attribute combined data, and then identify small faults based on these feature recognition parameters, which helps us understand the advantages and disadvantages of the model and further optimize the model.

[0103] The method for identifying small faults provided by an embodiment of the present invention analyzes seismic attributes using a preferred analysis method on seismic data from a study area to obtain corresponding fault characteristic information to be predicted, and then inputs the predicted fault characteristic information into a small fault identification model. This method can automatically identify small faults. When constructing the small fault identification model, the seismic attributes are analyzed using various preset analysis methods to obtain corresponding small fault characteristic information as sample data. Each of the preset analysis methods includes an analysis method combining at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume. This allows the sample data to contain at least two seismic attributes, resolving the problem of difficulty and low accuracy in identifying small faults in the prior art due to single seismic attribute analysis. Small fault identification is highly accurate and efficient. Furthermore, by comprehensively analyzing multiple seismic attributes and preferably selecting an analysis method for a combination of seismic attributes suitable for small fault identification, the structure, lithology, and fluid distribution of the stratum can be more accurately revealed, providing an important basis for identifying small faults, and achieving high accuracy in small fault identification.

[0104] Example 2

[0105] Based on the same inventive concept, see Figure 2 In an embodiment of the present invention, a method for training a small fault recognition model is provided, comprising the following steps:

[0106] Step S201: Based on the seismic data, various preset analysis methods are used to analyze the seismic attributes to obtain the characteristic information of the corresponding small faults as sample data, wherein each preset analysis method includes an analysis method of at least two attribute combinations of coherence body, ant tracking, hour window coherence body, diffusion filtering, and frequency division ant body.

[0107] Step S202: Train a pre-established deep learning model based on the sample data corresponding to each preset analysis method to obtain a small fault recognition training result.

[0108] Step S203: Evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result.

[0109] Step S204: Filter and obtain a preferred analysis method from various preset analysis methods according to the evaluation results corresponding to each evaluation parameter.

[0110] Step S205: Retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization.

[0111] Step S206: Repeat the training and optimization process of the optimal deep learning model until the model meets the preset evaluation conditions, thereby obtaining a small fault recognition model.

[0112] In the embodiment of the present invention, the specific implementation process of the training method of the small fault identification model can refer to the detailed description of the small fault identification method in the above embodiment 1, and will not be repeated here.

[0113] Example 3

[0114] Based on the same invention concept, Figure 11 In an embodiment of the present invention, a device for identifying small faults is provided, comprising:

[0115] Analysis module 101 is used to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each preset analysis method includes an analysis method combining at least two attributes: coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume;

[0116] The first training module 102 is used to train a pre-established deep learning model based on sample data corresponding to each preset analysis method to obtain a small fault identification training result;

[0117] The first evaluation module 103 is used to evaluate the small fault identification training results by presetting multiple evaluation parameters to obtain corresponding evaluation results;

[0118] The optimization module 104 is used to select a preferred analysis method from various preset analysis methods according to the evaluation results corresponding to each evaluation parameter;

[0119] The second training module 105 is used to retrain the optimal deep learning model corresponding to the optimal analysis method and perform model optimization;

[0120] A second evaluation module 106 is configured to evaluate the preferred deep learning model. If the preferred deep learning model does not meet the preset evaluation conditions, the second training module is enabled to re-execute the model training process. If the preferred deep learning model meets the preset evaluation conditions, a small fault recognition model is obtained.

[0121] An acquisition module 107 is configured to analyze seismic attributes of the acquired seismic data of the study area using an optimal analysis method to obtain corresponding fault characteristic information to be predicted;

[0122] The identification module 108 is used to input the characteristic information of the fault to be predicted into the small fault identification model to obtain the small fault identification result.

[0123] The implementation principle and technical effects of the device for identifying small faults provided in the embodiment of the present invention are similar to those of the first embodiment and will not be described in detail here.

[0124] Example 4

[0125] Based on the same inventive concept, an embodiment of the present invention provides a training device for a small fault identification model, comprising:

[0126] An analysis module is used to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each preset analysis method includes an analysis method combining at least two attributes: coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume;

[0127] The first training module is used to train a pre-established deep learning model based on sample data corresponding to each preset analysis method to obtain small fault identification training results;

[0128] The first evaluation module is used to evaluate the small fault identification training results by presetting multiple evaluation parameters to obtain corresponding evaluation results;

[0129] A selection module is used to select a preferred analysis method from various preset analysis methods based on the evaluation results corresponding to each evaluation parameter;

[0130] The second training module is used to retrain the optimal deep learning model corresponding to the optimal analysis method and perform model optimization;

[0131] The second evaluation module is used to evaluate the preferred deep learning model. If the preferred deep learning model does not meet the preset evaluation conditions, the second training module is enabled to re-execute the model training process. If the preferred deep learning model meets the preset evaluation conditions, a small fault recognition model is obtained.

[0132] The implementation principle and technical effects of the training device for the small fault identification model provided in the embodiment of the present invention are similar to those of the second embodiment and will not be described in detail here.

[0133] Example 5

[0134] Based on the same inventive concept, an embodiment of the present invention provides a computer-readable storage medium, which, when executed by a processor, implements the small fault identification method in embodiment 1 and / or the small fault identification model training method in embodiment 2.

[0135] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments, or may exist independently and not be incorporated into the device / apparatus. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to Embodiment 1 or Embodiment 2 of the present invention.

[0136] According to an embodiment of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, such as, but not limited to, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0137] Example 6

[0138] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, referring to Figure 12 , including a processor 111, a communication interface 112, a memory 113 and a communication bus 114, wherein the processor 111, the communication interface 112 and the memory 113 communicate with each other through the communication bus 114,

[0139] Memory 113, for storing computer programs;

[0140] The processor 111 is configured to implement the small fault identification method in the first embodiment and / or the small fault identification model training method in the second embodiment when executing the program stored in the memory 113 .

[0141] The implementation principle and technical effects of the electronic device provided by the embodiment of the present invention are similar to those of the aforementioned embodiment 1 or embodiment 2, and will not be repeated here.

[0142] The memory 113 can be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read-Only Memory), an EPROM, a hard disk, or a ROM. The memory 113 has storage space for program code for executing any of the method steps described above. For example, the storage space for program code can include individual program codes for implementing each step of the method described above. These program codes can be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards, or floppy disks. Such computer program products are typically portable or fixed storage units. The storage unit can have storage segments or storage spaces arranged similarly to the memory 113 in the electronic device described above. The program code can be compressed, for example, in a suitable form. Typically, the storage unit includes a program for executing the method steps according to Embodiment 1 or Embodiment 2 of the present invention, i.e., code that can be read by, for example, the processor 111. When executed by an electronic device, these codes cause the electronic device to execute the various steps of the method described in Embodiment 1 or Embodiment 2.

[0143] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0145] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0147] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying small faults, characterized in that: The following steps are involved: The seismic data of the study area are analyzed using the optimal analysis method to analyze the seismic attributes and obtain the corresponding fault characteristic information to be predicted; Input the characteristic information of the fault to be predicted into the small fault identification model to obtain a small fault identification result; wherein the small fault identification model is obtained by the following method: Based on the seismic data, various preset analysis methods are used to analyze seismic attributes to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes of coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume; According to the sample data corresponding to each of the preset analysis methods, a pre-established deep learning model is trained to obtain a small fault identification training result; Evaluate the small fault identification training result by presetting multiple evaluation parameters to obtain corresponding evaluation results; According to the evaluation results corresponding to each of the evaluation parameters, the preferred analysis method is screened from the preset analysis methods; Retraining the optimal deep learning model corresponding to the optimal analysis method and performing model optimization; The training and optimization process of the preferred deep learning model is repeated until the model meets the preset evaluation conditions, thereby obtaining the small fault identification model.

2. The method for identifying a small fault according to claim 1, wherein: After inputting the characteristic information of the fault to be predicted into the small fault identification model to obtain the small fault identification result, the method further includes: The small fault identification results are explained using AI visualization technology and spatially combined and displayed.

3. The method for identifying a small fault according to claim 1, wherein: Also includes: The seismic data is obtained by: Seismic wave data collected by a seismograph in a study area is obtained, and the seismic wave data is processed to obtain the seismic data.

4. The method for identifying a small fault according to claim 3, wherein: Also includes: The seismic data is preprocessed by constructing a guided filtering method until the seismic data meets the resolution requirement.

5. The method for identifying a small fault according to claim 1, wherein: The evaluation parameters include accuracy, recall, precision, F1 score and AUC-ROC; The evaluating the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result includes: The precision, recall rate, accuracy, F1 score and AUC-ROC are used to evaluate the small fault recognition training results to obtain corresponding evaluation results.

6. A training method for a small fault recognition model, characterized in that: The following steps are involved: Based on the seismic data, various preset analysis methods are used to analyze seismic attributes to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes of coherence volume, ant tracking, hour window coherence volume, diffusion filtering, and frequency division ant volume; According to the sample data corresponding to each of the preset analysis methods, a pre-established deep learning model is trained to obtain a small fault identification training result; Evaluate the small fault identification training result by presetting multiple evaluation parameters to obtain corresponding evaluation results; According to the evaluation results corresponding to each of the evaluation parameters, a preferred analysis method is screened from the preset analysis methods; Retraining the optimal deep learning model corresponding to the optimal analysis method and performing model optimization; The training and optimization process of the preferred deep learning model is repeated until the model meets the preset evaluation conditions, thereby obtaining the small fault identification model.

7. A device for identifying small faults, characterized in that: include: An analysis module is configured to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume; A first training module is used to train a pre-established deep learning model based on sample data corresponding to each of the preset analysis methods to obtain a small fault identification training result; A first evaluation module is used to evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result; A optimization module, configured to select the preferred analysis method from the preset analysis methods according to the evaluation results corresponding to each evaluation parameter; A second training module is used to retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization; a second evaluation module, configured to evaluate the preferred deep learning model; if the preferred deep learning model does not meet the preset evaluation conditions, causing the second training module to re-execute the model training process; if the preferred deep learning model meets the preset evaluation conditions, obtaining a small fault recognition model; An acquisition module is used to analyze seismic attributes of the acquired seismic data of the study area using an optimal analysis method to obtain corresponding fault feature information to be predicted; The identification module is used to input the characteristic information of the fault to be predicted into the small fault identification model to obtain a small fault identification result.

8. A training device for a small fault recognition model, characterized in that: include: An analysis module is configured to analyze seismic attributes based on seismic data using various preset analysis methods to obtain characteristic information of corresponding small faults as sample data; each of the preset analysis methods includes an analysis method combining at least two attributes: coherence volume, ant tracking, hourly window coherence volume, diffusion filtering, and frequency-divided ant volume; A first training module is used to train a pre-established deep learning model based on sample data corresponding to each of the preset analysis methods to obtain a small fault identification training result; A first evaluation module is used to evaluate the small fault identification training result by presetting a plurality of evaluation parameters to obtain a corresponding evaluation result; A optimization module, configured to select the preferred analysis method from the preset analysis methods according to the evaluation results corresponding to each evaluation parameter; A second training module is used to retrain the preferred deep learning model corresponding to the preferred analysis method and perform model optimization; The second evaluation module is used to evaluate the preferred deep learning model. If the preferred deep learning model does not meet the preset evaluation conditions, the second training module is enabled to re-execute the model training process. If the preferred deep learning model meets the preset evaluation conditions, a small fault recognition model is obtained.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for identifying a small fault according to any one of claims 1 to 5 and / or the method for training a small fault identification model according to claim 6 are implemented.

10. An electronic device, characterized in that: The processor, the communication interface, the memory and the communication bus are connected to each other via the communication bus. Memory for storing computer programs; The processor is configured to implement the method for identifying a small fault according to any one of claims 1 to 5 and / or the method for training a small fault identification model according to claim 6 when executing the program stored in the memory.

Images