A method for identifying the validity of ground microseismic picked events

Abstract

The present invention discloses a method for identifying the effectiveness of ground microseismic picked-up events, and the method includes the following steps: collecting the picked-up microseismic events after different microseismic business processes; annotating the microseismic events, converting the annotated microseismic event matrix into a signal feature matrix, and establishing a signal feature data set of microseismic events; defining the data sets used for training and validation; constructing a lightweight network model and training it using the signal characterization data of microseismic events as input; evaluating the model effect and evaluating the mean average precision value of the object detection model; and exporting and deploying the model for identifying the effectiveness of microseismic events. The method of the present invention can effectively realize the automatic identification of microseismic events, help improve the processing efficiency, and can also be used for quality control management of on-site supervision.

Description

Technical Field

[0001] The present invention relates to oil exploration and development technologies, and in particular to a method for identifying the effectiveness of ground microseismic picked-up events. Background Art

[0002] In the existing workflow, the identification of effective signal events adopts the long-short time window eigenvalue ratio method. Under the condition of high signal-to-noise ratio, by setting an appropriate threshold value, effective microseismic events can be automatically identified. However, ground microseismic data has the characteristic of low signal-to-noise ratio, and the long-short time window feature ratio method has a high misrecognition rate, which requires manual confirmation by engineers, bringing a large workload to the site. There is a need for a method for identifying the effectiveness of ground microseismic picked-up events to optimize the on-site business workflow and replace the manual confirmation link with a machine. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for identifying the effectiveness of ground microseismic picked-up events in view of the defects in the prior art.

[0004] The technical solution adopted by the present invention to solve its technical problems is as follows: A method for identifying the effectiveness of ground microseismic picked-up events, including:

[0005] 1) Collect the microseismic event data picked up in the ground microseismic business processing workflow;

[0006] 2) Preprocess the data obtained in step 1), label the microseismic events, convert the labeled microseismic event matrix into a signal feature matrix, and establish a signal feature data set of microseismic events;

[0007] Specifically as follows:

[0008] Combine the microseismic event matrix with the characteristics of the event time window, intercept the middle time window, and reduce the matrix dimension; label the microseismic events, perform SVD singular value decomposition on the labeled matrix, extract the eigenvalues for signal characterization, and further reduce the dimension;

[0009] 3) Divide the data set after the signal characterization of microseismic events into a training set, a validation set, and a test set;

[0010] 4) Construct a lightweight network model and use the signal characterization data of microseismic events as input for training. The network architecture of the lightweight network model is the lightweight network MobileNetV3;

[0011] 5) Evaluate the model effect and evaluate the mean average precision value of the object detection model;

[0012] 6) Export and deploy the model for identifying the effectiveness of microseismic events.

[0013] According to the above solution, in step 1), the microseismic event data is the microseismic event data detected by the long-short time window feature ratio algorithm.

[0014] According to the above solution, in step 2), the microseismic events are labeled, with the valid event labeled as class 1 and the invalid event labeled as class 0.

[0015] According to the above solution, in step 2), the signal feature data set of microseismic events is established as follows:

[0016] Let a three-dimensional microseismic signal data be S(x, y, t), where x = 1, 2, 3, …, Mx; y = 1, 2, 3, …, Ny; t = 1, 2, 3, …, Ti;

[0017] The size of the microseismic data is Mx * Ny * Ti. For a given sampling time t1, the data slice of the microseismic signal record is shown as follows:

[0018]

[0019] At a given time point t1, there is a matrix S(x, y, t1). Each row of the matrix is arranged into a Hankel matrix, and the data of its slice is shown as follows:

[0020]

[0021] where i = 1, 2, 3, …, Mx;

[0022] Ri is a Hankel matrix of size RV × RH, where, RH = Ny - RV;

[0023] For each Hankel matrix Ri corresponding to each row, a Hankel block matrix H is constructed.

[0024]

[0025] The block Hankel matrix H is as square or approximately square as possible, with a size of (RV × Lx) × (RH × Ly) order, where, Lx = Mx - Ly;

[0026] According to the SVD principle, the matrix H has the following representation:

[0027]

[0028] Suppose there is a matrix M = d + n, where M represents the noisy signal, d represents the valid signal, and n represents the noise. For each Hankel matrix after rearrangement of each slice, singular value decomposition is performed as follows:

[0029]

[0030] When reconstructing the signal, the contribution of the i-th eigen-signal is proportional to the i-th singular value. Since the singular values are arranged in decreasing order, the parts that contribute the most to the reconstruction of the microseismic event signal record are all included in the most principal eigenvalues. The selection and truncation of the singular values are as follows:

[0031]

[0032] Extract the eigenvalues for signal characterization to further reduce the dimension.

[0033] According to the above scheme, in step 2), the microseismic event data is a matrix with a dimension of 1031*2133. Then, in combination with the characteristics of the event time window, the middle time window is intercepted, and the matrix dimension is reduced to 300*2132; the microseismic events are labeled, and the labeled matrix is subjected to SVD singular value decomposition, and the eigenvalues are extracted for signal characterization, and the dimension is further reduced to 300*500.

[0034] According to the above scheme, in step 4), the Mobilenet v3 model is used to train the model with the microseismic event signal characterization as the input, the training parameters and the total number of epochs are set, and the learning rate is set to 0.001.

[0035] According to the above scheme, in step 5), the model effect is evaluated by evaluating the mean average precision value of the object detection model, and the best model at each stage is saved every 2000 epochs during the model training.

[0036] The beneficial effects produced by the present invention are:

[0037] The method of the present invention can realize the automatic identification of effective microseismic events, which helps to improve the processing efficiency and can also be used for the quality control management of on-site supervision. Brief Description of the Drawings

[0038] The present invention will be further described below in conjunction with the drawings and embodiments. In the drawings:

[0039] Figure 1 is the flowchart of the method of the embodiment of the present invention. Detailed Embodiments

[0040] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] As Figure 1As shown in the figure, a method for identifying the validity of ground microseismic pick-up events, the implementation process of this method includes:

[0042] 1) Collect the microseismic event data picked up in the ground microseismic business processing flow;

[0043] 2) Preprocess the data obtained in step 1), annotate the microseismic events, convert the annotated microseismic event matrix into a signal feature matrix, and establish a signal feature data set of microseismic events;

[0044] Specifically as follows:

[0045] Since the microseismic event recording signal after NMO has strong coherence of effective signals and weak coherence of random noise in the time - space domain. Assume that a three - dimensional microseismic signal data is S(x, y, t),

[0046] where x = 1, 2, 3, …, Mx; y = 1, 2, 3…Ny; t = 1, 2, 3, …, Ti; the size of the microseismic data is Mx * Ny * Ti; given the sampling time t1, the data slice of the microseismic signal record is shown as formula (1):

[0047]

[0048] At the given time point t1, there is a matrix S(x, y, t1), and each row of the matrix is arranged into a Hankel matrix. The data of its slice is shown as formula (2):

[0049]

[0050] R i is a Hankel matrix of size RV×RH, where, RH = Ny - RV.

[0051] For each Hankel matrix Ri corresponding to each row, construct a Hankel block matrix H, as shown in formula (3):

[0052]

[0053] The block Hankel matrix H is as square or approximately square as possible,

[0054] of order (RV×Lx)×(RH×Ly), where, Lx = Mx - Ly.

[0055] According to the SVD principle, the matrix H can be represented as follows

[0056]

[0057] Suppose there is a matrix \(M = d + n\), where \(M\) represents the noisy signal, \(d\) represents the valid signal, and \(n\) represents the noise. For each rearranged Hankel matrix of each slice, performing singular value decomposition gives:

[0058]

[0059] When reconstructing the signal, the contribution of the \(i\)-th eigen-signal is proportional to the \(i\)-th singular value. Since the singular values are arranged in decreasing order, the parts that contribute the most to the reconstruction of the microseismic event signal records are all contained in the most dominant eigenvalues. Selecting and truncating the singular values is the key step for MSSA to achieve signal characterization, as follows:

[0060]

[0061] The microseismic event data is a matrix with dimensions of \(1031\times2133\). Then, combined with the characteristics of the event time window, the middle time window is intercepted, and the matrix dimensions are reduced to \(300\times2132\); the microseismic events are labeled, and the labeled matrix is subjected to SVD (Singular Value Decomposition) to extract eigenvalues for signal characterization, further reducing the dimensions to \(300\times500\).

[0062] 3) Divide the dataset after signal characterization of microseismic events into a training set, a validation set, and a test set;

[0063] 4) Construct a lightweight network model and use the signal characterization data of microseismic events as input for training. The network architecture of the lightweight network model is the lightweight network MobileNetV3;

[0064] 5) Evaluate the model performance and evaluate the mean average precision value of the object detection model;

[0065] 6) Export and deploy the model for identifying the effectiveness of microseismic events.

[0066] In this embodiment, each step is specifically as follows:

[0067] S01. Collect microseismic events detected by the long-term average over short-term average (LTA / STA) algorithm in surface microseismic monitoring during fracturing construction;

[0068] Collecting microseismic events detected by the long-term average over short-term average (LTA / STA) algorithm in surface microseismic monitoring during fracturing construction can be data from one well or multiple wells;

[0069] S02. The microseismic event data is a matrix with dimensions of 1031 * 2031. Program to implement a microseismic event labeling tool, where the valid event label is class 1 and the invalid event label is class 0. Then convert the labeled matrix into an image. Combining the characteristics of the event time window, intercept the middle time window and reduce the image dimensions to 300 * 2031.

[0070] S03. Divide the labeled microseismic event images into training set, validation set, and test set. Use the random sampling method to divide the data into training set, validation set, and test set, and the ratio of the training set, validation set, and test set is 6:2:2.

[0071] S04. Modify the training configuration file, use the Mobilenet v3 model, train the model using the microseismic event images, set the training parameters, set the total number of epochs to 28000, and set the learning rate to 0.000125.

[0072] The training method can be methods such as stochastic gradient descent, Adam, etc. When the training reaches the convergence condition (the recognition rate exceeds the set value), terminate the training.

[0073] S05. Save the model every 2000 epochs during this training. Those named after the epoch number are all stage models. The model output after training is saved in the output / train folder. last is the model saved at the end of training, and model_best is the best model after each evaluation. Use the mean average precision value for evaluating the object detection model to evaluate the model effect. The larger the value, the better the model performance.

[0074] S06. Optimize the model, export and deploy the model for identifying the validity of microseismic events, load it into the ground microseismic monitoring software, and embed it into the ground microseismic data processing business process. It can also be used alone in combination with the business process.

[0075] The trained model is exported as a CPU and GPU model for identifying the validity of ground microseismic events that can be used alone or in third-party programs.

[0076] In summary, for the method for identifying the validity of ground microseismic pick-up events in this embodiment of the present invention, based on Mobilenet v3, the microseismic event data is labeled, image conversion is completed, model construction and training are carried out, solving the problem that the misrecognition rate of the long-short time window feature ratio method is relatively high due to the low signal-to-noise ratio of ground microseismic data, which requires manual confirmation by engineers and brings a large workload to the site.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

[0078] The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, a server, an App application store, etc., on which a computer program is stored, and when the program is executed by a processor, it implements a method for identifying the effectiveness of ground microseismic pick-up events in the method embodiments.

[0079] It should be pointed out that according to the needs of implementation, each step / component described in the present application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0080] It should be understood that those of ordinary skill in the art can make improvements or transformations according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for identifying the validity of ground microseismic picked-up events, characterized in that Including: 1) Collect the picked microseismic event data in the ground microseismic business processing flow; 2) Preprocess the data obtained in step 1), label the microseismic events, convert the labeled microseismic event matrix into a signal feature matrix, and establish a signal feature data set of microseismic events; Specifically as follows: Combine the microseismic event matrix with the characteristics of the event time window, intercept the middle time window, and reduce the matrix dimension; label the microseismic events, perform SVD singular value decomposition on the labeled matrix, extract eigenvalues for signal characterization, and further reduce the dimension; 3) Divide the data set after signal characterization of microseismic events into a training set, a validation set, and a test set; 4) Construct a lightweight network model and use the signal characterization data of microseismic events as input for training. The network architecture of the lightweight network model is the lightweight network MobileNetV3; 5) Evaluate the model effect and evaluate the mean average precision value of the object detection model; 6) Export and deploy the model for identifying the effectiveness of microseismic events.

2. The method for identifying the effectiveness of ground microseismic picked-up events according to claim 1, wherein In step 1), the microseismic event data is the microseismic event data detected by the long-short time window feature ratio algorithm.

3. The method for identifying the validity of ground microseismic pick-up events according to claim 1, wherein In step 2), label the microseismic events, with the valid event label being class 1 and the invalid event label being class 0.

4. The method for identifying the effectiveness of ground microseismic pick-up events according to claim 1, wherein In step 2), establish a signal feature data set of microseismic events, specifically as follows: Let a three-dimensional microseismic signal data be S(x, y, t), where x = 1, 2, 3, …, Mx; y = 1, 2, 3…Ny; t = 1, 2, 3, …, Ti; The size of the microseismic data is Mx*Ny*Ti. For a given sampling time t1, the data slice of the microseismic signal record is shown as follows: At the given time point t1, there is a matrix S(x, y, t1). Arrange each row of the matrix into a Hankel matrix, and the data of its slice is shown as follows: Where i = 1, 2, 3, …, Mx; Ri is a Hankel matrix of size RV×RH, where, RH = Ny - RV; For each Hankel matrix Ri corresponding to each row, construct a Hankel block matrix H, The matrix H is a square matrix or an approximate square matrix, with a size of (RV×Lx)×(RH×Ly), where, Lx = Mx - Ly; According to the SVD principle, the matrix H has the following representation: Suppose there is a matrix M = d + n, where M represents the noisy signal, d represents the valid signal, and n represents the noise. For each Hankel matrix after rearrangement of each slice, perform singular value decomposition: When reconstructing the signal, the contribution of the i-th eigen-signal is proportional to the i-th singular value. Since the singular values are arranged in descending order, the parts that contribute the most to the reconstruction of the microseismic event signal record are all included in the most principal eigenvalues. Select and truncate the singular values as follows: Extract eigenvalues for signal characterization and further reduce the dimension.

5. The method for identifying the effectiveness of ground microseismic picked-up events according to claim 4, characterized in that, In step 2), the microseismic event data is a matrix with a dimension of 1031*2133. Then, combine the characteristics of the event time window, intercept the middle time window, and reduce the matrix dimension to 300*2132; label the microseismic events, perform SVD singular value decomposition on the labeled matrix, extract eigenvalues for signal characterization, and further reduce the dimension to 300*500.

6. The method for identifying the validity of ground microseismic pick-up events according to claim 1, characterized in that, In step 4), the Mobilenet v3 model is used for model training with the microseismic event signal characterization as the input. The training parameters and the total number of epochs are set, and the learning rate is set to 0.

001.

7. The method for identifying the effectiveness of ground microseismic picked-up events according to claim 1, characterized in that In step 5), the model effect is evaluated by evaluating the mean average precision value of the object detection model, and the best model at each stage is saved every 2000 epochs during model training.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of a method for identifying the effectiveness of ground microseismic pick-up events according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, a method for identifying the effectiveness of ground microseismic pick-up events according to any one of claims 1 to 6 is implemented.

Images