A vertical well microseismic event automatic detection method based on space-time topology trajectory fitting, a computer device and a storage medium

Abstract

The application provides a vertical well microseismic event automatic detection method based on space-time topology trajectory fitting, a computer device and a storage medium. The method comprises the following steps: firstly, arranging and integrating channel data in time and space sequence according to collected data; then, performing self-adaptive decomposition and reconstruction on the arranged data by using empirical mode decomposition, so as to filter high-frequency components and part of middle and low-frequency components; secondly, picking all event points by using a classical short-time average or long-time average energy ratio algorithm; finally, performing secondary screening and fitting modeling on the event points by combining DBSCAN density clustering and RANSAC robust fitting, so as to obtain a fitting curve, and determining the continuous event time distribution in the well section direction by judgment.

Description

Technical Field

[0001] This invention belongs to the fields of microseismic monitoring, geophysical exploration and signal processing technology, and particularly relates to an automatic detection method, computer equipment and storage medium for vertical well microseismic events based on spatiotemporal topological trajectory fitting. Background Technology

[0002] Microseismic monitoring technology plays a crucial role in shale gas hydraulic fracturing monitoring, mine safety early warning, and geological disaster monitoring. However, in actual operating environments, the raw signals acquired often contain strong environmental background noise, resulting in a low signal-to-noise ratio (SNR) for microseismic signals. Traditional microseismic detection methods mostly rely on single-channel triggering or manual pickup. Single-channel triggering is prone to generating numerous false trigger points under low SNR conditions and struggles to distinguish random noise pulses from actual seismic waves.

[0003] Manual picking is extremely inefficient and cannot handle large-scale real-time monitoring data. Although there are some multi-channel joint detection methods, most of them have strict requirements on the geometric arrangement of the sensors and lack robust statistical processing mechanisms for discrete false trigger points generated by each channel, making it difficult to balance the false detection rate and the missed detection rate.

[0004] Therefore, automatically detecting microseismic signals generated by rock fractures in large-scale real-time monitoring data with low signal-to-noise ratios has become a technical challenge that urgently needs to be solved in the field of geological engineering. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an automatic detection method, computer equipment and storage medium for vertical well microseismic events based on spatiotemporal topology trajectory fitting. By converting multi-channel discrete event points into a global spatiotemporal point set and using trajectory fitting algorithms and their derivative variants for waveform extraction, microseismic signals in low signal-to-noise ratio environments can be detected automatically and accurately.

[0006] The present invention adopts the following technical solution: An automatic detection method for microseismic events in vertical wells based on spatiotemporal topological trajectory fitting includes: Step 1. Acquire real-time data: Use distributed or discrete signal sensing unit arrays to acquire time-series signals containing microseismic wave characteristics; Step 2. Establish a three-dimensional spatial model: Based on the spatial topology information of the signal sensing unit array in the preset coordinate system in Step 1, arrange and integrate the data of each channel to construct a spatiotemporal data matrix that reflects wave propagation; Step 3. Process the acquired real-time data: preprocess the spatiotemporal data matrix from Step 2, suppress environmental noise and enhance the effective signal components, and output the enhanced dataset to be analyzed; Step 4. Candidate event cluster detection: Based on the preset feature extraction criteria, detect and obtain all suspected event points in each channel of the augmented dataset in Step 3, extract and record their corresponding spatiotemporal coordinate parameters, and form a cluster of all candidate event points. Step 5. Trajectory Fitting and Waveform Recognition: Using trajectory fitting algorithms and their derivative variants, spatial clustering and curve reconstruction are performed on the candidate event point clusters mentioned in Step 4. By judging the goodness of fit, trajectory sequences that conform to the laws of wave kinematics are extracted, thereby completing the detection of microseismic waveforms.

[0007] Furthermore, the signal sensing unit array in step 1 includes a distributed optical fiber sensing system (DAS) and its improved versions, a piezoelectric detector, an accelerometer and a MEMS sensor, or any transducer capable of converting a continuous electrical signal or an optical signal through physical properties.

[0008] Furthermore, the preset coordinate system in step 2 includes Cartesian coordinates, polar coordinates, or spherical coordinates established with any geographical reference, the physical location of the sensing unit, or the preset seismic source estimation area as the origin. Spatial topology information includes the one-dimensional cable length, two-dimensional planar coordinates, or three-dimensional spatial vectors of the sensing unit.

[0009] Furthermore, step 3 preprocessing includes any one or a combination of linear / nonlinear transformation, time-frequency clipping, and noise reduction.

[0010] Among them, linear / nonlinear transformations include logarithmic transformation, exponential enhancement, Hilbert transformation, or normalization.

[0011] Time-frequency cutting: signal truncation based on energy threshold or subband division based on frequency range.

[0012] Noise reduction processing includes bandpass filtering, adaptive filtering, wavelet denoising, variational mode decomposition, or singular value decomposition.

[0013] The preset feature extraction criteria in step 4 are: Allows multiple candidate event points to exist within a single channel; All points that meet the initial value characteristics (including energy jumps, frequency changes, polarity reversals, or abnormal autocorrelation coefficients) are collected and stored in the database. The strongest event in a single channel is not used as the sole criterion. All possible false event points or manually preset points are retained for subsequent global weighted filtering.

[0014] Furthermore, the trajectory fitting algorithm in step 5 includes any one or a combination of the following categories: analytical and statistical fitting, transform domain feature extraction, nonparametric and clustering algorithms, and intelligent learning and optimization.

[0015] Analytical and statistical fitting algorithms include least squares (LS), weighted least squares (WLS), and derivative algorithms based on robust regression, including but not limited to RANSAC, PROSAC, MSAC, M-estimation, or Huber penalty function fitting.

[0016] Transform domain feature extraction class: includes Hough Transform, Generalized Hough Transform (GHT), Randomized Hough Transform (RHT), Radon Transform, and various projection deformation algorithms in the spatiotemporal domain.

[0017] Non-parametric and clustering algorithms: This category includes density-based clustering (DBSCAN), the OPTICS algorithm, mean shift, and probabilistic fitting algorithms based on Gaussian mixture models (GMM).

[0018] Intelligent learning and optimization category: includes neural network regression, support vector machine (SVM) regression, or curve fitting models optimized using heuristic search algorithms (such as genetic algorithms and particle swarm optimization).

[0019] Step 5 also includes a trajectory fitting process, which includes a weight-based iterative cleanup method, comprising: Initial weight assignment: Assign initial weights to each candidate event point based on its energy intensity or saliency. Spatial constraint screening: Calculate the algebraic or geometric distance from each candidate event point to the fitted trajectory; Dynamic weight update: Reduce the weight of isolated points that are far from the fitted curve, and increase the weight of point cloud clusters that conform to the continuous law of wave propagation. Convergence determination: Through multiple iterations, the goodness of fit or residual converges, and finally the detected microseismic waveform is composed of the event point sequence with high confidence weight.

[0020] In another embodiment of the present invention, a computer device is provided, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, performs the steps of the above-described method for automatic detection of microseismic events in vertical wells based on spatiotemporal topological trajectory fitting.

[0021] In another embodiment of the present invention, a computer storage medium is provided, which stores a computer program. When the computer program is executed by a processor, it performs the steps of the above-described method for automatic detection of microseismic events in vertical wells based on spatiotemporal topological trajectory fitting.

[0022] The beneficial effects of this invention are: This invention changes the excessive reliance on signal-to-noise ratio in the traditional single-channel triggering mode, enabling the system to automatically reconstruct the microseismic travel time curve from a massive number of scattered event points in monitoring environments with strong background noise and complex interference sources, by utilizing the physical consistency logic of wave propagation in the spatiotemporal distribution. This greatly improves the robustness and accuracy of waveform acquisition.

[0023] Meanwhile, the spatial topology mapping relationship constructed by this invention not only supports traditional discrete detectors, but is also compatible with modern monitoring methods such as distributed optical fiber sensing systems (DAS), eliminating the dependence on specific observation geometries.

[0024] Furthermore, the weight-based iterative mechanism can intelligently identify and suppress the interference of nonlinear random false triggering event points, realizing an automated processing flow from data acquisition to waveform detection for microseismic events. Attached Figure Description

[0025] Figure 1 This is a flowchart of the steps of the present invention.

[0026] Figure 2 This is a flowchart of the microseismic signal fitting process according to an embodiment of the present invention.

[0027] Figure 3 The following is a diagram showing the machine-automated identification results of microseismic signals according to an embodiment of the present invention. (a) is the original seismic data profile, showing the input original microseismic records; (b) is the data profile after normalization processing, showing the feature enhancement effect of the target event after preprocessing; and (c) is the result of automatic picking and high-precision fitting of microseismic events, showing the final detection effect of the method of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0029] like Figure 1 As shown, the present invention provides an automatic detection method for microseismic events in vertical wells based on spatiotemporal topological trajectory fitting, comprising: Real-time data acquisition is performed. A distributed fiber optic acoustic sensing system is used as the core signal sensing unit. Monitoring channels are divided at 2-meter intervals, with a total of n effective monitoring channels. The data acquisition frequency is 5000Hz, and the data from each channel is divided into 1.2-second intervals for subsequent processing.

[0030] Establish a three-dimensional spatial model. Using wellhead coordinates ( A three-dimensional spatial coordinate system is constructed with the origin as the reference point, and the coordinates of each detection channel are set sequentially from the wellhead downwards. Where i takes the values ​​1, 2, 3, ..., n. The data from each channel are arranged sequentially, and finally, an n×m dimensional array of data is obtained in chronological order.

[0031] like Figure 2 As shown, the above n×m dimensional array data is processed to suppress environmental noise and enhance the effective microseismic signal. To filter out high-frequency components and some mid-to-low-frequency components, this invention adaptively performs the above processing using empirical mode decomposition.

[0032] Microseismic picking was performed on the processed data. The STA / LTA algorithm was set to extract all suspected event points in each channel, and each event point was extracted and recorded. Its corresponding two-dimensional coordinate parameters ,all , constitute a set ,in For trigger time, This indicates the location of the fiber optic channel.

[0033] Finally, all suspected event points were screened and fitted. First, data points were clustered using DBSCAN, with a pre-defined neighborhood radius. And the minimum number of points, MinPts, then all data points are calculated according to... The relationship with the minimum number of points, MinPts, is automatically divided into three categories, among which... Neighborhood radius for each suspected event point The number of data points within is specifically calculated as follows: : 1. The core point (A) is the center of a high-density region within its neighborhood that contains a sufficient number of samples.

[0034] 2. Although the boundary point (C) does not satisfy the density condition, it is located in the neighborhood of the core point.

[0035] 3. Outliers (N) have a lower density and are identified as noise points.

[0036] This process can automatically identify spatiotemporally continuous clusters of real events and mark isolated noise events as outliers for removal. In actual processing, there may be multiple data clusters; further optimization and processing can be performed by selecting a few large clusters with more data points.

[0037] After completing DBSCAN density clustering, to further explore potential patterns in the data, the RANSAC algorithm is used for robust model fitting. A small number of data points are randomly sampled to estimate possible function models, with five preset parameters: 1. model: Fitting model (e.g., linear and quadratic curves, etc.).

[0038] 2.n: The minimum number of data sets used for fitting.

[0039] 3.k: The maximum number of traversals specified by the algorithm.

[0040] 4.t: The threshold for the degree of matching between data and model. Within the range of t, it is called inliers, and outside the range, it is called outliers.

[0041] 5.d: Indicates the minimum number of data sets suitable for the model.

[0042] Based on the image form of the seismic waves, a quadratic or higher-order curve model is selected for fitting. n, t, and d are set manually based on experience, and the number of iterations k is determined using a formula. Calculation, where This represents the probability that the RANSAC algorithm result is useful. Let be the probability that a data point is in the inliers set. Then, for each iteration of the model fitting, n data points are needed, and the probability that all of them are in the inliers set is... (Sampling with replacement probability), the probability of not being in the inliers set is... Therefore, the result of k iterations satisfies:

[0043] After setting appropriate parameters and iterating continuously, the model with the most interior points is selected as the optimal solution. Finally, the fitted curve obtained after the above processing is judged to determine whether it is a trajectory curve that conforms to the laws of wave kinematics. The curvature of the fitted curve is calculated, i.e. ,like If the value is less than a set value, it conforms to the laws of wave kinematics. A confidence level is then set based on the ratio of the number of inliers crossed by the fitted curve to the total number of data points, thus further filtering out waveform misjudgments.

[0044] The fitted curve obtained after the above processing can be determined as the continuous arrival time distribution of the event along the well section. Figure 3 This record shows the results of an earthquake waveform automatically identified by a machine. Figure 3 (a) and Figure 3In (b), each curve represents the time-domain record of a DAS channel within the event window. The waveform amplitude changes with the channel position. When the event occurs, there is a sudden change in amplitude at the channel cluster. Through waveform superposition, the data is processed and the entire waveform of the seismic wave is displayed in the heat map. Figure 3 In (c), the light blue lines correspond to the seismic waveforms obtained by fitting the picked points. The initial seismic values ​​picked by the machine are represented by purple diamonds, the gray points are the discrete points that have been filtered out, and the green points are the core points that participate in the fitting.

[0045] On the other hand, embodiments of the present invention also provide a microseismic signal identification system, comprising: Data acquisition module: used to acquire raw fluctuation signals and send data containing information such as timestamp, channel number, sampling rate and amplitude to the modeling module; Modeling module: The physical location information of the channels obtained from the data acquisition module is used to establish a reference coordinate system and map the channel topology. Then, the topology and coordinate system parameter results are sent to the calculation module, along with the cached data. The computation module comprises a preprocessing module, a feature detection module, and a trajectory fitting module. These three sub-modules sequentially or cyclically call the data transmitted from the modeling module, ultimately performing tasks such as extracting candidate point clusters, reconstructing the function of the fluctuating trajectory, and calculating the fitted data. This process then sends the fitting results to the output module.

[0046] Output module: Visualizes the fitted data from the calculation module and outputs the detected waveform time-travel features and event recognition parameters.

[0047] Another invention, according to an embodiment of the present invention, also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it performs the above-described method for automatic detection of vertical well microseismic events based on spatiotemporal topological trajectory fitting.

[0048] On the other hand, a computer storage medium is also provided, which stores a computer program. When the computer program is executed by a processor, it performs the above-mentioned method for automatic detection of microseismic events in vertical wells based on spatiotemporal topological trajectory fitting.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic detection method for microseismic events in vertical wells based on spatiotemporal topological trajectory fitting, characterized in that, include: Step 1. Obtain real-time data: By utilizing distributed or discrete arrays of signal sensing units, time-series signals containing microseismic wave characteristics can be acquired. Step 2. Establish a three-dimensional spatial model: Based on the spatial topology information of the signal sensing unit array in the preset coordinate system in step 1, the data of each channel are arranged and integrated to construct a spatiotemporal data matrix that reflects wave propagation. Step 3. Process the acquired real-time data: The spatiotemporal data matrix from step 2 is preprocessed to suppress environmental noise and enhance effective signal components, outputting the enhanced dataset to be analyzed. Step 4. Candidate event cluster detection: Based on the preset feature extraction criteria, all suspected event points in each channel of the augmented dataset in step 3 are detected and obtained, and their corresponding spatiotemporal coordinate parameters are extracted and recorded to form a cluster of all candidate event points. Step 5. Trajectory Fitting and Waveform Recognition: Using trajectory fitting algorithms and their derivative variants, spatial clustering and curve reconstruction are performed on the candidate event point clusters mentioned in step 4. By judging the goodness of fit and extracting trajectory sequences that conform to the laws of wave kinematics, the detection of microseismic waveforms is completed.

2. The method according to claim 1, characterized in that, Step 1 The signal sensing unit array includes a distributed fiber optic sensing system, a piezoelectric detector, an accelerometer, and a MEMS sensor.

3. The method according to claim 1, characterized in that, In step 2, the preset coordinate system includes a Cartesian coordinate system, a polar coordinate system, or a spherical coordinate system established with any geographical reference, the physical location of the sensing unit, or the preset seismic source estimation area as the origin; the spatial topology information includes the one-dimensional cable length, two-dimensional planar coordinates, or three-dimensional spatial vector of the sensing unit.

4. The method according to claim 1, characterized in that, Step 3 preprocessing includes any one or a combination of linear or nonlinear transformation, time-frequency clipping, and noise reduction.

5. The method according to claim 1, characterized in that, Step 5: Trajectory fitting algorithms include any one of the following categories or combinations: analytical and statistical fitting, transform domain feature extraction, nonparametric and clustering algorithms, and intelligent learning and optimization.

6. The method according to claim 1, characterized in that, The preset feature extraction criteria in step 4 are: Allows multiple candidate event points to exist within a single channel; All points that meet the initial value characteristics are collected and stored in the database. The strongest event in a single channel is not used as the sole criterion. All false event points or manually preset points are retained for subsequent global weighted filtering.

7. The method according to claim 1, characterized in that, Step 5 also includes a trajectory fitting process, which includes a weight-based iterative cleanup method, comprising:

501. Initial weight assignment: Assign initial weights to each candidate event point based on its energy intensity or salience.

502. Spatial Constraint Screening: Calculate the algebraic or geometric distance from each candidate event point to the fitted trajectory; 503. Dynamic weight update: Reduce the weight of isolated points that are far from the fitted curve, and increase the weight of point cloud clusters that conform to the continuity law of wave propagation; 504. Convergence Judgment: Through multiple iterations, the goodness of fit or residual converges, and finally the detected microseismic waveform is composed of the event point sequence with high confidence weight.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it performs the steps of the automatic detection method for vertical well microseismic events based on spatiotemporal topological trajectory fitting as described in any one of claims 1-7.

9. A computer storage medium storing a computer program, wherein when the computer program is executed by a processor, it is characterized in that, The steps of the automatic detection method for vertical well microseismic events based on spatiotemporal topological trajectory fitting according to any one of claims 1-7 are performed.

Images