Early warning method based on microseismic energy accumulation tendency

Abstract

The invention provides an early warning method based on microseismic energy accumulation tendency, and belongs to the field of seismic data processing. The method comprises the following steps: acquiring a historical micro-seismic log set of an earthquake monitoring sensor array in a coal mine area; performing staged processing in combination with a timestamp to obtain a micro-seismic energy characteristic sub-sequence set; executing a first-order difference operation, and constructing a difference energy characteristic subsequence set; performing trend correlation analysis by using a space-time guiding module to obtain a trend feature set; constructing an energy accumulation tendency analyzer based on the trend feature set; and obtaining a real-time micro-seismic log, carrying out abnormity identification by combining with an analyzer, and triggering an early warning instruction. The technical problem that in the prior art, the dynamic change trend of the micro-seismic energy cannot be accurately captured, so that the early warning accuracy of the energy accumulation is insufficient is solved, and the technical effect of improving the early warning accuracy and timeliness of the micro-seismic energy accumulation through differential energy feature analysis and space-time trend association is achieved.

Description

Technical Field

[0001] The present invention relates to the field of seismic data processing, and in particular to an early warning method based on microseismic energy accumulation tendency. Background Art

[0002] During coal mining, microseismic events often occur due to the redistribution of geological stress and changes in rock mass structure. Abnormal accumulation of microseismic energy is often a significant precursor to coal mine disasters and accidents. Therefore, effective monitoring and early warning of microseismic energy accumulation are crucial for ensuring safe coal mine production.

[0003] At present, existing microseismic monitoring and early warning technologies are mainly based on the analysis of static characteristics such as the energy amplitude and frequency of microseismic events. For example, abnormal conditions are judged by setting energy thresholds or event frequency thresholds, or statistical methods are used to analyze the distribution characteristics of microseismic energy. These methods can identify microseismic anomalies to a certain extent, but there are obvious technical defects. Specifically, existing technologies mainly focus on the absolute numerical characteristics of microseismic energy, while ignoring the dynamic process of energy change. The accumulation of microseismic energy is a gradual dynamic process, and its changing trend and rate of change can often better reflect the proximity of disasters than absolute values. However, existing methods lack the ability to accurately capture the dynamic change trend of microseismic energy and cannot effectively identify the tendency characteristics of energy accumulation, resulting in insufficient early warning accuracy, prone to false alarms or missed alarms, and difficult to meet the actual needs of coal mine safety production. Summary of the Invention

[0004] The present invention aims to solve the technical problem that the existing technology cannot accurately capture the dynamic change trend of microseismic energy, resulting in insufficient accuracy of energy accumulation warning, and provides a warning method based on the tendency of microseismic energy accumulation to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] In a first aspect, the present invention provides an early warning method based on microseismic energy accumulation tendency, comprising: obtaining a set of historical microseismic logs collected by an earthquake monitoring sensor array deployed in a target coal mine area within a historical time window, wherein each historical microseismic log includes a log timestamp; in combination with the log timestamp, processing the historical microseismic log set in stages to obtain a set of historical microseismic energy characteristic subsequences; performing first-order difference operations on the historical microseismic energy characteristic subsequences to construct a set of historical differential energy characteristic subsequences; performing trend correlation analysis on the historical microseismic energy characteristic subsequences and the historical differential energy characteristic subsequences using a time-space guidance module to obtain a set of historical microseismic energy trend characteristics and a set of historical differential energy trend characteristics; constructing an energy accumulation tendency analyzer based on the historical microseismic energy trend characteristics and the historical differential energy trend characteristics; obtaining a real-time microseismic log sequence set obtained by real-time monitoring of the earthquake monitoring sensor array, performing anomaly identification in combination with the energy accumulation tendency analyzer, and triggering a microseismic energy accumulation early warning instruction.

[0007] Optionally, the historical microseismic log set is processed in stages in combination with the log timestamp to obtain a historical microseismic energy feature subsequence set, including: obtaining a pre-built feature extractor to extract energy features from the historical microseismic log set to obtain a historical microseismic energy feature set; serializing the historical microseismic energy feature set according to the log timestamp to obtain a historical microseismic energy feature sequence; slicing the historical microseismic energy feature sequence according to a preset sliding window to determine a historical microseismic energy feature subsequence set.

[0008] Optionally, the historical microseismic energy characteristic sequence is sliced ​​according to a preset sliding window to determine a set of historical microseismic energy characteristic subsequences, including: slicing the historical microseismic energy characteristic sequence according to a preset sliding window to obtain an initial set of historical microseismic energy characteristic subsequences; performing adjacent fusion analysis on the initial set of historical microseismic energy characteristic subsequences to obtain the set of historical microseismic energy characteristic subsequences.

[0009] Optionally, a time-space guidance module is used to perform trend correlation analysis on the historical microseismic energy feature subsequence set and the historical differential energy feature subsequence set, respectively, to obtain a historical microseismic energy trend feature set and a historical differential energy trend feature set, including: based on a preset time scale and a sample time-guided training set, supervised training is performed on a framework constructed based on a feedforward neural network to obtain a time-guided submodule; based on a preset space scale and a sample space-guided training set, supervised training is performed on a framework constructed based on a feedforward neural network to obtain a space-guided submodule; and the time-guided submodule and the space-guided submodule are connected in parallel to obtain the time-space guidance module.

[0010] Optionally, the method also includes: respectively using the time-oriented submodule and the space-oriented submodule to extract time and space features of the historical microseismic energy feature sequence and the historical differential energy feature sequence to obtain historical microseismic energy time trend features, historical microseismic energy space trend features, historical differential energy time trend features, and historical differential energy space trend features; performing trend correlation analysis on the historical microseismic energy time trend features and the historical microseismic energy space trend features to obtain historical microseismic energy trend features; performing trend correlation analysis on the historical differential energy time trend features and the historical differential energy space trend features to obtain historical differential energy trend features.

[0011] Optionally, a trend correlation analysis is performed on the historical microseismic energy time trend characteristics and the historical microseismic energy space trend characteristics to obtain the historical microseismic energy trend characteristics, including: calculating the feature similarity of the historical microseismic energy time trend characteristics and the historical microseismic energy space trend characteristics to obtain a feature similarity set; normalizing the feature similarity set and matrixing the processing results to obtain a historical microseismic energy time adjacency matrix; and convolving the historical microseismic energy time adjacency matrix and the historical microseismic energy space trend characteristics to obtain the historical microseismic energy trend characteristics.

[0012] Optionally, an energy accumulation tendency analyzer is constructed based on the historical microseismic energy trend feature set and the historical differential energy trend feature set, including: taking each historical microseismic energy trend feature and historical differential energy trend feature in the historical microseismic energy trend feature set and the historical differential energy trend feature set as a group of positive sample pairs to obtain a positive sample pair set; mapping the historical microseismic energy trend feature set and the historical differential energy trend feature set into a historical microseismic energy trend feature probability distribution and a historical differential energy trend feature probability distribution, respectively; performing a symmetric KL divergence calculation on the historical microseismic energy trend feature probability distribution and the historical differential energy trend feature probability distribution to determine a consistency loss function; and performing supervised calculation on a framework constructed based on a convolutional neural network using the positive sample pair set as a comparative representation learning input, and constructing an energy accumulation tendency analyzer by minimizing the consistency loss function.

[0013] Optionally, the historical microseismic energy trend feature set and the historical differential energy trend feature set are mapped into the historical microseismic energy trend feature probability distribution and the historical differential energy trend feature probability distribution, respectively, including: normalizing and softmax mapping the historical microseismic energy trend feature set to obtain the historical microseismic energy trend feature probability distribution; normalizing and softmax mapping the historical differential energy trend feature set to obtain the historical differential energy trend feature probability distribution.

[0014] Optionally, a real-time microseismic log sequence set obtained by real-time monitoring of the seismic monitoring sensor array is obtained, and anomaly identification is performed in combination with the energy accumulation tendency analyzer to trigger a microseismic energy accumulation warning instruction, including: performing energy feature extraction on the real-time microseismic log sequence set to determine a real-time microseismic energy feature sequence set; performing a first-order difference operation on the real-time microseismic energy feature sequence set to determine a real-time differential energy feature sequence set; using a time-space guidance module to perform trend correlation analysis on the real-time microseismic energy feature sequence set and the real-time differential energy feature sequence set respectively to determine a real-time microseismic energy trend feature set and a real-time differential energy trend feature set; using the energy accumulation tendency analyzer to perform consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set to determine a consistency loss value set; parsing the consistency loss value set, and if there is an anomaly, triggering a microseismic energy accumulation warning instruction.

[0015] Optionally, the energy accumulation tendency analyzer is used to perform consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set to determine the consistency loss value set, including: obtaining the monitoring coordinates corresponding to the consistency loss values ​​greater than or equal to the preset consistency loss value threshold in the consistency loss value set to obtain an abnormal monitoring coordinate set; dividing the abnormal monitoring coordinate set into neighboring areas according to the preset monitoring coordinate neighbor threshold to obtain K divided neighboring areas, where K is a positive integer; traversing and counting the number of abnormal monitoring coordinates in the K divided neighboring areas, and placing the statistical results in the areas of the K divided neighboring areas to obtain abnormal coefficients of the K divided neighboring areas; when any one of the K divided neighboring area abnormal coefficients is greater than or equal to the preset divided neighboring area abnormal coefficient threshold, a microseismic energy accumulation warning instruction is triggered.

[0016] The beneficial effects of the present invention are:

[0017] Acquire a set of historical microseismic logs collected by the seismic monitoring sensor array deployed in the target coal mine area within the historical time window, wherein each historical microseismic log includes a log timestamp, thereby establishing a complete historical microseismic data foundation and providing data support for subsequent energy characteristic analysis; combine the log timestamps to process the historical microseismic log set in stages to obtain a set of historical microseismic energy characteristic subsequences, thereby segmenting the continuous historical data according to the time dimension, facilitating the capture of energy change laws in different periods; perform first-order difference operations on the historical microseismic energy characteristic subsequence sets respectively, and construct a set of historical differential energy characteristic subsequences, thereby converting absolute energy values ​​into relative changes, highlighting the dynamic characteristics and change rates of energy changes; use the time-space guidance module to analyze the energy characteristics of the historical microseismic energy characteristic subsequences. Perform trend correlation analysis on the historical microseismic energy characteristic subsequence set and the historical differential energy characteristic subsequence set respectively to obtain the historical microseismic energy trend characteristic set and the historical differential energy trend characteristic set, explore the correlation law of energy change from the two dimensions of time and space, and identify the trend characteristics of energy accumulation; construct an energy accumulation tendency analyzer based on the historical microseismic energy trend characteristic set and the historical differential energy trend characteristic set to provide an analysis and decision-making basis for real-time monitoring; obtain the real-time microseismic log sequence set obtained by real-time monitoring of the seismic monitoring sensor array, combine it with the energy accumulation tendency analyzer to perform anomaly identification, trigger the microseismic energy accumulation warning instruction, realize real-time analysis and judgment of the current microseismic state, and issue a warning in time when an abnormal energy accumulation tendency is detected.

[0018] Through the above technical solution, by capturing the dynamic change characteristics of microseismic energy and combining it with the time-space guidance module to perform trend correlation analysis, an early warning mechanism based on energy accumulation tendency is constructed, which effectively solves the technical problem that the existing technology cannot accurately capture the dynamic change trend of microseismic energy, and achieves the technical effect of improving the accuracy and timeliness of microseismic energy accumulation early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic flow chart of an early warning method based on microseismic energy accumulation tendency provided by the present invention;

[0020] Figure 2 A schematic diagram of the process of triggering the microseismic energy accumulation early warning instruction provided by the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0023] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0024] like Figure 1 As shown, an embodiment of the present invention provides an early warning method based on the tendency of microseismic energy accumulation, including:

[0025] S1. Obtain a set of historical microseismic logs collected by an earthquake monitoring sensor array deployed in a target coal mine area within a historical time window, wherein each historical microseismic log includes a log timestamp.

[0026] Specifically, the target coal mine area refers to a specific coal mining area that requires early warning of microseismic energy accumulation. This area includes the working face that is being mined or about to be mined and its surrounding impact range, specifically covering the coal mining working face, heading tunnel, goaf, and surrounding rock areas that may be affected by mining disturbances. A seismic monitoring sensor array is pre-deployed in the target coal mine area. The seismic monitoring sensor array consists of a number of microseismic monitoring sensors, which are distributed and installed in different locations underground in the coal mine, such as key areas such as main tunnels, near the working face, and at the edge of the goaf. Each sensor is equipped with a high-precision vibration detection element and a data processing unit, which can monitor the microseismic signals generated by the coal and rock mass during the mining process in real time, and convert the detected vibration data into digital signals for recording and transmission.

[0027] The historical time window refers to the specific time range used to obtain historical microseismic logs. The length of this time window is determined based on the coal mine's geological conditions, mining progress, and historical microseismic activity patterns. For example, for an active working face, the historical time window can be set from three months before mining to the current time to ensure sufficient historical microseismic logs are obtained for subsequent analysis. During this historical time window, the seismic monitoring sensor array operates continuously, and data recording is automatically triggered when a vibration signal exceeding a preset threshold is detected. Each historical microseismic log is a complete digital record of a single microseismic event, including the log timestamp of the microseismic event. The log timestamp is a required field in each historical microseismic log, accurately recording the exact time of the microseismic event. It is typically formatted as "year-month-day hour:minute:second.millisecond," such as "2024-03-15 14:25:36.123." In addition to the log timestamp, each historical microseismic log also contains information such as the magnitude, source location coordinates (x, y, z), energy released, and the sensor number. These historical microseismic logs are stored in chronological order to form a historical microseismic log collection, which provides a data basis for subsequent microseismic energy feature extraction and accumulation tendency analysis.

[0028] S2. Process the historical microseismic log set in stages based on the log timestamps to obtain a set of historical microseismic energy characteristic subsequences.

[0029] Specifically, the acquired historical microseismic log collection is organized and segmented according to the time dimension to extract characteristic subsequences that reflect the variation patterns of microseismic energy. The segmented processing involves arranging the historical microseismic log collection in a time series according to the log timestamps. The entire historical time window is then divided into multiple consecutive time stages according to preset time period lengths, with each time stage corresponding to a historical microseismic energy characteristic subsequence.

[0030] First, the historical microseismic log collection is sorted chronologically based on the log timestamps within each log, forming an ordered sequence in ascending order. Next, a preset stage duration is set. This stage duration can be adjusted based on the periodic characteristics of microseismic activity and early warning requirements, with typical values ​​being 1, 6, or 24 hours. The entire historical time window is then segmented into several consecutive, non-overlapping time periods, using this stage duration as an interval. All historical microseismic logs within each time period constitute a processing stage.

[0031] For each processing stage of the historical microseismic log, microseismic energy-related parameters are extracted, including information such as magnitude, energy release, and focal depth. Statistical features within that stage are then calculated, such as average energy, maximum energy, energy fluctuation variance, and microseismic event frequency. These statistical features are combined in chronological order to form a subsequence that reflects the temporal variation characteristics of historical microseismic energy, namely, a historical microseismic energy feature subsequence. Because the entire historical time window is divided into multiple processing stages, multiple historical microseismic energy feature subsequences are ultimately obtained, forming a collection of historical microseismic energy feature subsequences, providing structured data for subsequent energy accumulation tendency analysis.

[0032] S3. Perform first-order difference operations on the historical microseismic energy characteristic subsequence sets respectively to construct a historical differential energy characteristic subsequence set.

[0033] Specifically, a first-order difference operation is used to capture the changing trends and fluctuation patterns of historical microseismic energy characteristics, thereby constructing a set of historical differential energy characteristic subsequences that can reflect the tendency of energy accumulation. The first-order difference operation refers to the subtraction of two adjacent data points in the historical microseismic energy characteristic subsequence set to obtain the data change rate information.

[0034] For each historical microseismic energy characteristic subsequence in the set of historical microseismic energy characteristic subsequences, the difference in microseismic energy characteristic values ​​between two adjacent time points is calculated according to the chronological order of the data points in the subsequence. For example, if a historical microseismic energy characteristic subsequence contains a series of energy data arranged in chronological order, such as the energy value of the first time period, the energy value of the second time period, and the energy value of the third time period, the first-order difference operation process for this subsequence is as follows: subtract the energy value of the first time period from the energy value of the second time period to obtain the first difference value; subtract the energy value of the second time period from the energy value of the third time period to obtain the second difference value; and so on, until the energy differences between all adjacent time periods are calculated.

[0035] Through first-order difference calculations, the original absolute energy values ​​are converted into energy rate of change information. Positive values ​​indicate an upward trend in microseismic energy over adjacent time periods, while negative values ​​indicate a downward trend. The magnitude of the difference reflects the severity of the energy change. This differential feature effectively eliminates the influence of the energy baseline value and highlights the dynamic characteristics of microseismic energy changes, particularly identifying abnormal patterns of rapid energy accumulation or sudden release.

[0036] After performing the above first-order difference operation on each historical microseismic energy characteristic subsequence in the set, the corresponding historical differential energy characteristic subsequence is obtained. All these historical differential energy characteristic subsequences are combined to form the historical differential energy characteristic subsequence set. This historical differential energy characteristic subsequence set complements the original historical microseismic energy characteristic subsequence set, providing rich characteristic information for subsequent spatiotemporal guidance analysis and energy accumulation tendency assessment.

[0037] S4. Use a spatiotemporal guidance module to perform trend correlation analysis on the historical microseismic energy feature subsequence set and the historical differential energy feature subsequence set, respectively, to obtain a historical microseismic energy trend feature set and a historical differential energy trend feature set.

[0038] Specifically, through the time-space guidance module, an in-depth trend correlation analysis is conducted on historical microseismic data (historical microseismic energy characteristic subsequence set and historical differential energy characteristic subsequence set), and the changing patterns and correlation characteristics of microseismic energy are explored from the perspectives of time and space dimensions, thereby obtaining a trend feature set (historical microseismic energy trend feature set and historical differential energy trend feature set) that can reflect the tendency of microseismic energy accumulation.

[0039] Among them, the spatiotemporal guidance module is an intelligent analysis framework built based on a feedforward neural network, which includes two parallel working components: a time guidance submodule and a space guidance submodule.

[0040] The time-oriented submodule is specifically designed to analyze the evolution and trend characteristics of microseismic energy signatures over time series. This submodule receives as input a set of historical microseismic energy signature subsequences or a set of historical differential energy signature subsequences. Using its internal neural network structure, it identifies temporal patterns of energy variation, such as periodic fluctuations, gradual growth, and sudden changes. This submodule is capable of capturing energy variation patterns across different time scales, including both short-term transient changes and long-term cumulative trends.

[0041] The spatial guidance submodule is specifically designed to analyze the spatial correlation and propagation characteristics of microseismic energy signatures. Combining the source location information of microseismic events, this submodule analyzes energy correlations between different spatial locations and identifies spatial patterns of energy accumulation, such as localized concentration, regional diffusion, and directional propagation. The spatial guidance submodule can also detect energy interactions and influence patterns between adjacent regions.

[0042] During the trend correlation analysis process, the time-oriented and space-oriented submodules operate in parallel, extracting features in the temporal and spatial dimensions, respectively, from the same set of input data. For the historical microseismic energy feature subsequences, the two submodules extract their temporal and spatial trend features, respectively. Then, through a feature fusion mechanism, they integrate the spatiotemporal information and ultimately output a historical microseismic energy trend feature set. Similarly, for the historical differential energy feature subsequences, the same spatiotemporal analysis process is followed to output a historical differential energy trend feature set. These two trend feature sets comprehensively reflect the complex variations in microseismic energy across the spatiotemporal dimensions, providing feature input for subsequent assessments of energy accumulation trends.

[0043] S5. Constructing an energy accumulation tendency analyzer based on the historical microseismic energy trend feature set and the historical differential energy trend feature set.

[0044] Specifically, based on a set of historical microseismic energy trend features and a set of historical differential energy trend features, an intelligent analyzer capable of determining the tendency of microseismic energy accumulation, known as an energy accumulation tendency analyzer, was constructed. This energy accumulation tendency analyzer, a deep learning model based on contrastive representation learning and consistency loss calculation, is used to identify the consistency relationship between microseismic energy signatures and differential energy signatures, and based on this, determine whether there is an energy accumulation tendency.

[0045] The construction process of this energy accumulation tendency analyzer includes training sample construction, probability distribution mapping, loss function design, and model training. First, the feature data from the corresponding time periods in the historical microseismic energy trend feature set and the historical differential energy trend feature set are paired. Each pair of historical microseismic energy trend features and historical differential energy trend features is considered a positive sample pair, representing the consistency relationship that should be maintained between the two features under normal conditions. This pairing method constructs a set of positive sample pairs for model training. Next, probability distribution conversion is performed on the historical microseismic energy trend feature set and the historical differential energy trend feature set. Through normalization and Softmax function mapping, the original trend feature values ​​are converted into probability distributions, eliminating the influence of different feature dimensions and making subsequent consistency comparisons more accurate and stable. A consistency loss function is then designed to measure the degree of difference between the two trend features. This consistency loss function, based on the symmetric KL divergence calculation method, quantifies the degree of consistency between the two trend features by comparing the similarity between the probability distributions of the historical microseismic energy trend features and the probability distributions of the historical differential energy trend features. When the two features are highly consistent, the loss value is small; when there are significant differences between the two features, the loss value is large. Subsequently, a deep learning framework based on a convolutional neural network was used, with a set of positive sample pairs as training input, to optimize the network parameters through supervised learning. During the training process, the network learns to minimize the consistency loss function, enabling the model to accurately identify the consistency pattern between the microseismic energy characteristics and the differential energy characteristics under normal conditions. After training is completed, the model becomes an energy accumulation tendency analyzer. In practical applications, when the consistency loss value calculated from the input real-time microseismic data exceeds the preset threshold, the energy accumulation tendency analyzer determines the presence of an energy accumulation tendency and outputs a warning signal.

[0046] S6. Acquire a real-time microseismic log sequence set obtained by the real-time monitoring of the seismic monitoring sensor array, perform anomaly identification in combination with the energy accumulation tendency analyzer, and trigger a microseismic energy accumulation early warning instruction.

[0047] Specifically, the constructed energy accumulation tendency analyzer is used to conduct online analysis of real-time monitoring data to realize the real-time early warning function of microseismic energy accumulation.

[0048] The seismic monitoring sensor array continues to monitor in real time, continuously collecting information on microseismic activity within the target coal mine area. Similar to the historical data collection process, each microseismic event generated during real-time monitoring is recorded as a real-time microseismic log, containing key information such as the time of occurrence, magnitude, focal location, and energy release. These real-time microseismic logs are organized chronologically to form a real-time microseismic log sequence. This real-time microseismic log sequence reflects the latest status and trends of microseismic activity within the current period.

[0049] Subsequently, the same feature extraction and processing process as for historical data was performed on the acquired real-time microseismic log sequence set. First, energy features were extracted from the real-time microseismic log sequence set to obtain a real-time microseismic energy feature sequence set. Then, a first-order difference operation was performed on this energy feature sequence set to construct a real-time differential energy feature sequence set. Next, trend correlation analysis was performed on the two types of real-time feature sequences using the trained spatiotemporal guidance module, respectively, to obtain a real-time microseismic energy trend feature set and a real-time differential energy trend feature set.

[0050] During the anomaly identification phase, the real-time microseismic energy trend feature set and the real-time differential energy trend feature set are input into the energy accumulation tendency analyzer for consistency loss value calculation. The energy accumulation tendency analyzer calculates the degree of difference between the current real-time features based on the consistency pattern learned in the training phase and outputs a consistency loss value set. The consistency loss value set is parsed, and loss values ​​greater than or equal to the preset consistency loss value threshold are identified as anomalies. When an anomaly is detected, the spatial distribution characteristics of the anomaly are further analyzed, and the presence of a regional energy accumulation tendency is determined through neighboring area division and anomaly coefficient calculation. Once the existence of an energy accumulation tendency is confirmed, the microseismic energy accumulation warning instruction is immediately triggered, and a warning message is sent to the coal mine safety management personnel so that appropriate safety protection measures can be taken in a timely manner.

[0051] Furthermore, the historical microseismic log set is processed in stages in combination with the log timestamps to obtain a set of historical microseismic energy characteristic subsequences, including:

[0052] S21, obtaining a pre-built feature extractor to perform energy feature extraction on the historical microseismic log set to obtain a historical microseismic energy feature set;

[0053] S22, serializing the historical microseismic energy feature set according to the log timestamp to obtain a historical microseismic energy feature sequence;

[0054] S23 , slicing the historical microseismic energy characteristic sequence according to a preset sliding window to determine a set of historical microseismic energy characteristic subsequences.

[0055] In a preferred embodiment, first, an energy feature extraction operation is performed on the historical microseismic log set through a pre-built feature extractor. The feature extractor is a data processing module specifically used to extract key energy parameters from the original microseismic logs, and is pre-designed and configured according to the physical characteristics of microseisms and the needs of energy accumulation analysis. The feature extractor extracts multidimensional energy-related features from each historical microseismic log, mainly including key parameters such as the magnitude of the microseismic event, the amount of energy released, the coordinates of the hypocenter location, the depth of the hypocenter, the duration, and the frequency characteristics. The magnitude reflects the intensity level of the microseismic event; the amount of energy released represents the total energy value released by the microseismic event; the coordinates of the hypocenter location identify the three-dimensional spatial location of the microseismic event; and the depth of the hypocenter describes the location of the microseismic event in the vertical direction. By performing feature extraction operations on all logs in the historical microseismic log set, a historical microseismic energy feature set containing rich energy information is obtained.

[0056] Then, based on the log timestamp corresponding to each historical microseismic energy feature, the historical microseismic energy feature set is time-serialized. Specifically, all historical microseismic energy features are rearranged in the order of the log timestamps to ensure that they are organized in the chronological order of the microseismic events. During the serialization process, multiple microseismic energy features with the same or similar timestamps are merged and processed to calculate the statistical characteristic values ​​within the time period, such as the average magnitude, total energy release, event frequency, etc. Through this serialization process, the originally discretely distributed historical microseismic energy features are converted into a continuous sequence arranged in chronological order, forming a historical microseismic energy feature sequence. This historical microseismic energy feature sequence can clearly reflect the evolution law and change trend of microseismic energy over time.

[0057] Subsequently, a preset sliding window is used to slice the historical microseismic energy characteristic sequence, dividing the continuous time series into multiple subsequences with a fixed time span. Among them, the preset sliding window refers to the time window used to split the sequence. Its length is set according to the periodic characteristics of microseismic activity and the timeliness requirements of warning. Typical sliding window lengths are 2 hours, 6 hours, 12 hours or 24 hours. During the sliding window slicing process, starting from the starting position of the historical microseismic energy characteristic sequence, data segments of equal length are intercepted in chronological order based on the preset sliding window length. Each data segment constitutes a historical microseismic energy characteristic subsequence. In order to ensure the continuity and integrity of the data, a certain degree of overlap can be set between adjacent sliding windows, usually 10% to 50% of the sliding window length. Through the sliding window slicing process, the historical microseismic energy characteristic sequence is decomposed into multiple subsequences of consistent length and continuous time. These subsequences constitute the historical microseismic energy characteristic subsequence set, which provides a data input format for subsequent differential operations and trend analysis.

[0058] Furthermore, the historical microseismic energy characteristic sequence is sliced ​​according to a preset sliding window to determine a set of historical microseismic energy characteristic subsequences, including:

[0059] S231, slicing the historical microseismic energy characteristic sequence according to a preset sliding window to obtain an initial historical microseismic energy characteristic subsequence set;

[0060] S232: Perform adjacent fusion analysis on the initial historical microseismic energy characteristic subsequence set to obtain the historical microseismic energy characteristic subsequence set.

[0061] In a preferred embodiment, first, the historical microseismic energy characteristic sequence is sliced ​​according to a preset sliding window to obtain an initial set of historical microseismic energy characteristic subsequences. Specifically, the time length of the preset sliding window is used as the slicing unit, and starting from the starting time point of the historical microseismic energy characteristic sequence, data segments are sequentially intercepted at fixed time intervals. Each sliding window covers the same time span, ensuring that the subsequences obtained by slicing have a uniform time length and data structure. The slicing process adopts an equal interval division method, that is, there is no overlap or gap between adjacent sliding windows, and the entire historical microseismic energy characteristic sequence is completely divided into several continuous time periods. Through this initial slicing process, an initial set of historical microseismic energy characteristic subsequences is obtained, and each subsequence in the set corresponds to a specific time window and contains all the microseismic energy characteristic data within the time window.

[0062] Next, a neighboring fusion analysis is performed on the initial set of historical microseismic energy signature subsequences to identify and merge adjacent subsequences with similar characteristics, reducing data redundancy and improving the efficiency of subsequent analysis. During the neighboring fusion analysis, the feature similarity between each pair of temporally adjacent subsequences in the initial set of historical microseismic energy signature subsequences is first calculated. The similarity is quantified by comparing the differences in statistical characteristic parameters, such as average energy value, energy fluctuation variance, and peak frequency, between the two temporally adjacent subsequences. When the feature similarity between two adjacent subsequences exceeds a preset similarity threshold, the two subsequences are considered to represent essentially identical microseismic activity patterns and meet the fusion criteria. The fusion operation merges adjacent similar subsequences along the temporal dimension to form a new subsequence with a longer time span. The fused new subsequence contains all the data information from the original two subsequences, and its feature parameters are recalculated using weighted averaging or cumulative calculation. A traversal neighboring fusion analysis is performed on the entire set of initial historical microseismic energy signature subsequences to identify all adjacent subsequence pairs that meet the fusion criteria and then perform the fusion operation. This adjacent fusion process effectively smooths out any minor fluctuations or noise that might have existed, enhancing the overall trend characteristics of the data while also reducing the total number of subsequences. After adjacent fusion analysis, an optimized set of historical microseismic energy signature subsequences is obtained, which preserves the key information of the original data while improving data quality and analytical applicability.

[0063] Furthermore, a trend correlation analysis is performed on the historical microseismic energy characteristic subsequence set and the historical differential energy characteristic subsequence set using a spatiotemporal guidance module to obtain a historical microseismic energy trend characteristic set and a historical differential energy trend characteristic set, including:

[0064] S41, based on a preset time scale and a sample time-oriented training set, performing supervised training on a framework constructed based on a feedforward neural network to obtain a time-oriented submodule;

[0065] S42, based on the preset spatial scale and the sample spatial orientation training set, supervised training is performed on the framework constructed based on the feedforward neural network to obtain a spatial orientation submodule;

[0066] S43: Connect the time-oriented submodule and the space-oriented submodule in parallel to obtain the space-time-oriented module.

[0067] In a preferred embodiment, a time-oriented submodule is first constructed, specifically for analyzing the temporal evolution of microseismic energy. Specifically, a preset time scale is first determined. The preset time scale refers to the standard time granularity used for temporal feature analysis and is set based on the temporal periodicity of microseismic activity. Typical time scales include hourly, daily, and weekly granularities. Simultaneously, a sample time-oriented training set is constructed based on the determined preset time scale. This sample time-oriented training set contains a large number of labeled microseismic energy time series samples and their corresponding time trend labels, which are used to train a neural network to identify different temporal variation patterns, such as increasing energy trends, periodic fluctuations, and sudden changes. This time-oriented submodule utilizes a deep learning framework based on a feedforward neural network, comprising multiple fully connected layers and activation function layers, capable of learning complex nonlinear time mapping relationships. During supervised training, the microseismic energy time series data in the sample time-oriented training set is used as input, and the corresponding time trend labels are used as the desired output. The network parameters are continuously adjusted through a backpropagation algorithm, enabling the network to accurately identify and predict the temporal evolution trends of microseismic energy. The training process uses batch gradient descent optimization, setting an appropriate learning rate and number of training rounds until the network converges and reaches the preset performance indicators. After training, the neural network framework becomes a time-oriented submodule, capable of extracting temporal trend features from microseismic energy sequences.

[0068] At the same time, a spatial guidance submodule is constructed to analyze the spatial distribution of microseismic energy. Specifically, a preset spatial scale is first determined. This scale refers to the standard spatial granularity used for spatial feature analysis. It is set based on the spatial extent of the target coal mine area and the spatial distribution characteristics of microseismic events. Typical spatial scales include meters, decads, and hundreds of meters, with different spatial resolutions. Based on the preset spatial scale, a sample spatial guidance training set is constructed. This sample spatial guidance training set contains a large number of labeled samples of the spatial distribution of microseismic events and their corresponding spatial trend labels. This is used to train a neural network to identify different spatial distribution patterns, such as local aggregation, regional diffusion, and directional propagation. The spatial guidance submodule also utilizes a deep learning framework based on a feedforward neural network, but its network structure is specifically optimized for the characteristics of spatial data, capable of processing multidimensional data containing spatial information such as source location coordinates and spatial distance relationships. During supervised training, the spatial distribution data of microseismic events in the sample spatial guidance training set is used as input, and the corresponding spatial trend labels are used as the desired output. Network parameters are optimized through a similar backpropagation training process. After training, the neural network framework becomes a spatially oriented submodule, capable of extracting spatial trend features from the spatial distribution of microseismic events.

[0069] Subsequently, the trained time-oriented and space-oriented submodules are connected in parallel to form a complete space-time-oriented module. Parallel connection means that the two submodules operate simultaneously in the data processing flow, processing the time and spatial dimensions of the same set of input data, respectively, and independently extracting corresponding trend features. Specifically, when a set of historical microseismic energy feature subsequences or a set of historical differential energy feature subsequences is input into the space-time-oriented module, the time-oriented submodule focuses on analyzing the temporal evolution characteristics of these sequences, while the space-oriented submodule focuses on analyzing the spatial distribution characteristics of the corresponding microseismic events. The outputs of the two submodules are integrated through a feature fusion mechanism to form a comprehensive space-time trend feature. The feature fusion process uses a cascaded splicing approach to merge the temporal and spatial trend features into a unified feature vector that simultaneously contains information on the variation patterns of microseismic energy in both time and space. Through parallel connection and feature fusion, the space-time-oriented module can comprehensively capture the complex spatiotemporal evolution patterns of microseismic energy, providing high-quality feature support for accurately identifying energy accumulation tendencies.

[0070] Furthermore, the embodiment of the present application also includes:

[0071] S44, respectively using the time-oriented submodule and the space-oriented submodule to extract spatiotemporal features from the historical microseismic energy feature sequence and the historical differential energy feature sequence to obtain historical microseismic energy time trend features, historical microseismic energy spatial trend features, historical differential energy time trend features, and historical differential energy spatial trend features;

[0072] S45, performing trend correlation analysis on the historical microseismic energy time trend characteristics and the historical microseismic energy spatial trend characteristics to obtain historical microseismic energy trend characteristics;

[0073] S46. Perform trend correlation analysis on the historical differential energy time trend characteristics and the historical differential energy space trend characteristics to obtain the historical differential energy trend characteristics.

[0074] In a preferred embodiment, the constructed time-oriented submodule and space-oriented submodule are first used to perform feature extraction operations on the historical microseismic energy signature sequence and the historical differential energy signature sequence, respectively, in the temporal and spatial dimensions. Specifically, the time-oriented submodule receives the historical microseismic energy signature sequence as input, analyzes the temporal evolution of the sequence through its internal feedforward neural network structure, identifies the temporal patterns and trend characteristics of energy changes, and outputs the historical microseismic energy time trend characteristics. This historical microseismic energy time trend characteristic reflects the temporal variation of microseismic energy, including time-related characteristic information such as energy growth rate, fluctuation period, and mutation time.

[0075] Meanwhile, the spatial guidance submodule receives the same historical microseismic energy signature sequence as input, but focuses on analyzing the spatial distribution characteristics of the corresponding microseismic events. By analyzing spatial information such as the source location coordinates, spatial concentration, and distribution directionality of the microseismic events, this submodule extracts the spatial distribution patterns and propagation characteristics of microseismic energy, and outputs the spatial trend characteristics of historical microseismic energy. This historical microseismic energy spatial trend characteristic reflects the spatial distribution pattern of microseismic energy, including spatially relevant characteristic information such as energy concentration areas, diffusion directions, and spatial correlations.

[0076] Using the same processing flow, the time-oriented submodule and the space-oriented submodule extract features from the historical differential energy feature sequence, outputting the historical differential energy time trend feature and the historical differential energy spatial trend feature, respectively. The historical differential energy time trend feature reflects the evolution of the microseismic energy change rate in the time dimension, while the historical differential energy spatial trend feature reflects the distribution characteristics of the energy change rate in the spatial dimension.

[0077] Subsequently, trend correlation analysis was performed on the temporal and spatial trend features of historical microseismic energy extracted from the same data source. The goal was to organically integrate the temporal and spatial feature information to obtain an overall trend feature that comprehensively reflects the spatiotemporal evolution of microseismic energy. During the trend correlation analysis, the correlation strength between the temporal and spatial trend features was first calculated to identify their complementarity and consistency in describing microseismic energy variations. A feature weighting mechanism was then employed to assign fusion weights to the temporal and spatial trend features based on their contribution to energy accumulation prediction. Finally, the temporal and spatial trend features were combined through weighted fusion to form a unified feature vector. This feature vector incorporates both the temporal evolution and spatial distribution information of microseismic energy, comprehensively describing the spatiotemporal variation of microseismic energy. After trend correlation analysis, a historical microseismic energy trend feature was obtained, which comprehensively reflects the complex variation pattern of microseismic energy in both temporal and spatial dimensions.

[0078] The same trend correlation analysis method as step S45 is used to fuse the historical differential energy time trend feature and the historical differential energy space trend feature to obtain the historical differential energy trend feature. Since the differential energy feature reflects the rate of change information of microseismic energy, its trend correlation analysis pays more attention to the dynamic characteristics and mutation characteristics of energy change. By calculating the degree of correlation between the two differential trend features, the collaborative pattern of energy change rate in time and space dimensions is identified. After similar feature weight allocation and weighted fusion processing, the historical differential energy trend feature is obtained. This feature comprehensively reflects the spatiotemporal distribution law of microseismic energy change rate, and can effectively identify abnormal patterns of rapid energy accumulation or sudden release. The historical differential energy trend feature and the historical microseismic energy trend feature complement each other, providing more comprehensive and accurate feature input for subsequent energy accumulation tendency analysis.

[0079] Furthermore, trend correlation analysis is performed on the historical microseismic energy temporal trend characteristics and the historical microseismic energy spatial trend characteristics to obtain historical microseismic energy trend characteristics, including:

[0080] S451, calculating the feature similarity of the historical microseismic energy time trend feature and the historical microseismic energy spatial trend feature to obtain a feature similarity set;

[0081] S452, normalizing the feature similarity set and constructing the processing result into a matrix to obtain a historical microseismic energy-time adjacency matrix;

[0082] S453 . Convolve the historical microseismic energy time adjacency matrix and the historical microseismic energy spatial trend feature to obtain the historical microseismic energy trend feature.

[0083] In a preferred embodiment, first, the correlation between the two features is established by calculating the similarity between the historical microseismic energy time trend feature and the historical microseismic energy space trend feature. Specifically, the historical microseismic energy time trend feature and the historical microseismic energy space trend feature are respectively used as feature vectors of two different dimensions, and the similarity between the two is calculated by the inner product operation. The inner product operation can quantify the correlation between the two feature vectors in the numerical distribution. When the numerical change trends of the two feature vectors are consistent, the inner product value is large, indicating a high similarity; when the numerical change trends of the two feature vectors are opposite or unrelated, the inner product value is small, indicating a low similarity. The similarity of the historical microseismic energy time trend feature and the corresponding historical microseismic energy space trend feature in all time periods is calculated one by one to obtain the degree of correlation between the spatiotemporal features in different time periods, and form a feature similarity set. The feature similarity set comprehensively describes the correlation pattern of microseismic energy in the time dimension and the spatial dimension, and provides a quantitative correlation basis for subsequent feature fusion.

[0084] Subsequently, the feature similarity set was normalized and matrixed to obtain a temporal adjacency matrix for historical microseismic energy. Normalization involves scaling all values ​​in the feature similarity set to a standard range of zero to one, eliminating magnitude differences between different similarity values ​​and ensuring numerical stability for subsequent matrix operations. The normalized similarity values ​​were further processed using a Softmax function, converting each similarity value into a probability between zero and one, with the sum of all similarity values ​​equal to one. During the matrix construction process, the normalized feature similarity values ​​were organized into a two-dimensional matrix structure based on chronological order and feature correspondence. The rows and columns of this matrix correspond to different time periods or feature dimensions, respectively. The value of each element in the matrix represents the strength of the correlation between the temporal trend feature and the spatial trend feature at the corresponding row and column position. The constructed temporal adjacency matrix for historical microseismic energy reflects the global correlation between the temporal and spatial features of microseismic energy. Matrix elements with larger values ​​indicate stronger correlations between the corresponding features, while elements with smaller values ​​indicate weaker correlations.

[0085] Subsequently, a convolutional neural network was used to perform a convolution operation on the temporal adjacency matrix of historical microseismic energy and the spatial trend features of historical microseismic energy, achieving deep fusion of spatiotemporal features. Convolution is a feature fusion method based on convolutional neural networks. It extracts local correlation patterns between features through a sliding convolution kernel and combines these local patterns into a global feature representation. During the convolution fusion process, the temporal adjacency matrix of historical microseismic energy serves as a weight matrix to guide the weight distribution of feature fusion; the spatial trend features of historical microseismic energy serve as input features, providing the spatial information that needs to be enhanced. The convolution operation achieves weighted feature fusion based on correlation strength by performing element-by-element multiplication and accumulation operations on the adjacency matrix and the spatial trend features. Specifically, spatial trend features corresponding to elements with larger values ​​in the adjacency matrix receive higher weights and occupy a more prominent position in the fusion result, while features corresponding to elements with smaller values ​​in the adjacency matrix are correspondingly weakened. Through convolution fusion processing based on the adjacency matrix, temporal trend information can be effectively fused into spatial trend features, obtaining a comprehensive feature representation that simultaneously incorporates temporal evolution patterns and spatial distribution characteristics. The final output of the historical microseismic energy trend feature not only retains the core information of the original spatial trend feature, but also incorporates the interactive information from the temporal trend feature, achieving effective integration of multi-scale information and avoiding the information redundancy problem that may be caused by simple feature splicing.

[0086] Furthermore, an energy accumulation tendency analyzer is constructed based on the historical microseismic energy trend feature set and the historical differential energy trend feature set, including:

[0087] S51, taking each historical microseismic energy trend feature and each historical differential energy trend feature in the historical microseismic energy trend feature set and the historical differential energy trend feature set as a set of positive sample pairs to obtain a positive sample pair set;

[0088] S52, mapping the historical microseismic energy trend feature set and the historical differential energy trend feature set into a historical microseismic energy trend feature probability distribution and a historical differential energy trend feature probability distribution, respectively;

[0089] S53, performing a symmetric KL divergence calculation on the historical microseismic energy trend characteristic probability distribution and the historical differential energy trend characteristic probability distribution to determine a consistency loss function;

[0090] S54. By using the positive sample pair set as a comparative representation learning input, supervised calculation is performed on the framework constructed based on the convolutional neural network, and an energy accumulation tendency analyzer is constructed by minimizing the consistency loss function.

[0091] In a preferred embodiment, first, a set of positive sample pairs for comparative representation learning is constructed by pairing and combining. Specifically, the feature data corresponding to the same time period in the historical microseismic energy trend feature set and the historical differential energy trend feature set are paired one by one, and each historical microseismic energy trend feature and the historical differential energy trend feature of its corresponding time period form a set of positive sample pairs. This pairing method is based on the following basis: under normal microseismic activity, the absolute value feature of microseismic energy and its rate of change feature should maintain a certain internal consistency relationship, that is, the two should reflect the same microseismic activity pattern and energy evolution trend. Through traversal processing, corresponding positive sample pairs are established for each time period. These positive sample pairs represent the consistency pattern that should be maintained between the microseismic energy trend feature and the differential energy trend feature under normal conditions. All positive sample pairs are combined together to form a positive sample pair set, which provides standard training samples for subsequent comparative representation learning, so that the energy accumulation tendency analyzer can learn to identify the consistency relationship pattern between the two features under normal conditions.

[0092] Subsequently, the historical microseismic energy trend feature set and the historical differential energy trend feature set were converted into corresponding probability distribution forms for subsequent consistency quantitative analysis. The probability distribution mapping process includes normalization and Softmax function mapping. First, all feature values ​​in the historical microseismic energy trend feature set are normalized, uniformly scaling feature values ​​of different magnitudes to the same numerical range to eliminate the impact of dimensional differences on subsequent calculations. Then, the Softmax function is used to perform a probability mapping transformation on the normalized feature values, converting each feature value into a probability value between zero and one, ensuring that the sum of all probability values ​​within the same feature vector is equal to one. Using the same processing flow, the historical differential energy trend feature set is subjected to probability distribution mapping to obtain the historical differential energy trend feature probability distribution. Through this probability distribution mapping process, the original trend feature data is converted into a standardized probability distribution form, providing a unified data format for subsequent consistency loss calculations and making comparisons between different types of features more fair and accurate.

[0093] Subsequently, a consistency loss function was designed based on the symmetric KL divergence calculation method to quantify the degree of difference between the probability distribution of historical microseismic energy trend features and the probability distribution of historical differential energy trend features. Symmetric KL divergence is a mathematical metric that measures the similarity between two probability distributions. Its value reflects the degree of difference between the two distributions: when the two probability distributions are completely consistent, the symmetric KL divergence value is zero; when the two probability distributions differ significantly, the symmetric KL divergence value is large. By calculating the symmetric KL divergence between the probability distribution of the microseismic energy trend feature and the probability distribution of the differential energy trend feature in each positive sample pair, a loss value reflecting the degree of consistency between the two features is obtained. This loss value serves as the output of the consistency loss function and is used to guide the subsequent model training process. The design goal of the consistency loss function is to ensure that under normal conditions, the microseismic energy trend feature and the differential energy trend feature maintain high consistency, and the corresponding loss value should be small; under abnormal conditions, the consistency between the two features is broken, and the corresponding loss value should be significantly increased.

[0094] Subsequently, a deep learning framework based on a convolutional neural network was trained using a contrastive representation learning method. An energy accumulation tendency analyzer was constructed by minimizing a consistency loss function. Contrastive representation learning is a machine learning method that learns data representations by comparing similarities and differences between samples. It is suitable for learning consistent relationship patterns between data. During the training process, a set of positive sample pairs is used as input data. The microseismic energy trend features and differential energy trend features in each positive sample pair are input into the convolutional neural network framework for feature encoding and representation learning. Through multiple layers of convolution operations and nonlinear transformations, the network learns to extract deep representations of the two features and calculates the consistency loss between the two features. The training goal is to minimize the consistency loss function for positive sample pairs by adjusting the network parameters, thereby enabling the network to accurately identify consistent patterns between the two features under normal conditions. After sufficient supervised training, the deep learning framework becomes the energy accumulation tendency analyzer. In the actual deployment and application phase, the energy accumulation tendency analyzer receives the real-time microseismic energy trend features and differential energy trend features to be detected as input and calculates the consistency loss between them. When the calculated consistency loss value exceeds a preset threshold, the analyzer determines that the current microseismic activity has deviated from the normal consistency pattern and is prone to energy accumulation, and outputs a corresponding warning result. Conversely, when the consistency loss value is within the normal range, the analyzer determines that the current state is normal and there is no significant risk of energy accumulation.

[0095] Furthermore, the historical microseismic energy trend feature set and the historical differential energy trend feature set are mapped into a historical microseismic energy trend feature probability distribution and a historical differential energy trend feature probability distribution, respectively, including:

[0096] S521, normalizing and softmax mapping the historical microseismic energy trend feature set to obtain a probability distribution of historical microseismic energy trend features;

[0097] S522 : Normalize and softmax map the historical differential energy trend feature set to obtain a historical differential energy trend feature probability distribution.

[0098] In a preferred embodiment, first, the historical microseismic energy trend feature set is converted into a probability distribution. Specifically, first, all feature data in the historical microseismic energy trend feature set are normalized, and the maximum and minimum values ​​of all values ​​in the historical microseismic energy trend feature set are calculated. Then, the maximum and minimum normalization method is used to linearly scale each feature value according to its relative position in the overall value range, so that all feature values ​​are mapped to a standard interval of zero to one. The normalization process can eliminate the numerical magnitude differences between different feature dimensions and ensure the numerical stability of subsequent processing. After the normalization process is completed, the processing result is subjected to a softmax mapping operation. Softmax mapping is a mathematical transformation method that converts any real vector into a probability distribution. The basic principle is to perform an exponential function transformation on each element in the vector, and then divide the transformation result by the sum of the exponential transformation results of all elements to obtain a probability distribution with a sum of one. Through softmax mapping, the normalized historical microseismic energy trend features are converted into a probability form, in which each feature value corresponds to a probability value between zero and one, and the sum of all probability values ​​in the same feature vector is strictly equal to one. After normalization and softmax mapping processing, the probability distribution of historical microseismic energy trend characteristics is obtained. This probability distribution of historical microseismic energy trend characteristics maintains the relative relationship of the original feature data and has the standard mathematical properties of probability distribution.

[0099] At the same time, the same processing flow as step S521 is used to convert the historical differential energy trend feature set into a probability distribution. First, normalization processing is performed on all feature data in the historical differential energy trend feature set, and the maximum and minimum values ​​of the historical differential energy trend feature set are calculated. Then, the same maximum and minimum value normalization method is used to linearly scale all values ​​of the differential energy trend feature to the standard range of zero to one. Next, a softmax mapping transformation is performed on the normalized differential energy trend feature data to convert each feature vector into a corresponding probability distribution form. Since the differential energy trend feature reflects the rate of change information of microseismic energy, its value may contain positive and negative values. The normalization process can effectively handle this bidirectional numerical characteristic, and the softmax mapping ensures that the final probability distribution has standard mathematical properties. After the same normalization and softmax mapping processing, the probability distribution of the historical differential energy trend feature is obtained. This historical differential energy trend probability distribution shares the same mathematical format and properties as the historical microseismic energy trend probability distribution. Both are expressed in a standardized probability form, providing uniform and numerically stable input data for subsequent symmetric KL divergence calculations and consistency analysis. Through symmetric probability distribution mapping, the absolute value and rate-of-change characteristics of microseismic energy are converted into directly comparable probability distributions, enabling accurate quantitative consistency analysis between the two distinct characteristics.

[0100] Further, such as Figure 2 As shown, obtaining a real-time microseismic log sequence set obtained by real-time monitoring of the seismic monitoring sensor array, combining the energy accumulation tendency analyzer to perform anomaly identification, and triggering a microseismic energy accumulation early warning instruction includes:

[0101] S61, performing feature extraction on the real-time microseismic log sequence set to determine a real-time microseismic energy feature sequence set;

[0102] S62, performing a first-order difference operation on the real-time microseismic energy characteristic sequence set to determine a real-time differential energy characteristic sequence set;

[0103] S63, using a spatiotemporal guidance module to perform trend correlation analysis on the real-time microseismic energy feature sequence set and the real-time differential energy feature sequence set, respectively, to determine a real-time microseismic energy trend feature set and a real-time differential energy trend feature set;

[0104] S64, using the energy accumulation tendency analyzer to perform consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set to determine a consistency loss value set;

[0105] S65: Analyze the consistency loss value set, and if an anomaly is found, trigger a microseismic energy accumulation warning instruction.

[0106] In a preferred embodiment, feature extraction is first performed on the real-time microseismic log sequence set obtained through real-time monitoring by the seismic monitoring sensor array to determine a real-time microseismic energy signature sequence set. The processing flow for this step is identical to the feature extraction process for historical data, but the processing target is real-time microseismic monitoring data. Specifically, the same feature extractor as in step S21 is used to extract energy-related parameters from each microseismic log in the real-time microseismic log sequence set, including key characteristic information such as the magnitude, energy release, focal location coordinates, focal depth, and duration of the microseismic event. Then, using the same serialization processing method as in step S22, the extracted energy features are arranged in chronological order based on the timestamp information of the real-time microseismic logs to form a real-time microseismic energy signature sequence. Next, using the same sliding window slicing processing method as in step S23, the real-time microseismic energy signature sequence is segmented according to a preset sliding window length, and adjacent fusion analysis is performed to ultimately obtain a real-time microseismic energy signature sequence set. This real-time microseismic energy signature sequence set reflects the real-time changes in microseismic energy during the current period, providing a data foundation for subsequent real-time analysis.

[0107] Subsequently, the real-time microseismic energy feature sequence set is processed using the same first-order difference operation method as step S3. For each feature sequence in the real-time microseismic energy feature sequence set, the difference in the microseismic energy characteristic values ​​between two adjacent time points is calculated in chronological order to obtain differential information reflecting the current microseismic energy change rate. Through the first-order difference operation, the dynamic change trend of microseismic energy can be captured in real time, and in particular, abnormal change patterns of rapid energy accumulation or sudden release can be promptly identified. After processing is completed, a real-time differential energy feature sequence set is obtained. This real-time differential energy feature sequence set complements the real-time microseismic energy feature sequence set, providing more comprehensive real-time microseismic activity feature information.

[0108] Next, using the trained spatiotemporal guidance module, trend correlation analysis is performed on the real-time microseismic energy feature sequence set and the real-time differential energy feature sequence set. The processing method of this step is the same as step S4, but the processing object is real-time data. The time-oriented submodule and the space-oriented submodule in the spatiotemporal guidance module work in parallel, performing in-depth analysis of the real-time feature data from the time dimension and the space dimension respectively, and extracting the spatiotemporal evolution law and trend characteristics of the current microseismic activity. Through the same feature extraction and fusion process as steps S44 to S46, the real-time microseismic energy time trend characteristics, real-time microseismic energy space trend characteristics, real-time differential energy time trend characteristics and real-time differential energy space trend characteristics are obtained. Then, through trend correlation analysis, the real-time microseismic energy trend feature set and the real-time differential energy trend feature set are finally obtained. These two trend feature sets comprehensively reflect the complex change laws of the current microseismic activity in the spatiotemporal dimension.

[0109] Subsequently, the constructed energy accumulation tendency analyzer was used to perform consistency loss analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set. The energy accumulation tendency analyzer receives two types of real-time trend features as input and, based on the consistency judgment model learned during the training phase, calculates the degree of consistency between the current real-time microseismic energy trend feature and the real-time differential energy trend feature. Specifically, the real-time trend features are converted into a probability distribution, and the symmetric KL divergence between the two probability distributions is calculated to obtain a loss value reflecting the degree of consistency of the current state. The same consistency loss value calculation is performed on the real-time trend features of all time periods to obtain a set of consistency loss values. Each loss value in this set corresponds to a specific time period and spatial region, reflecting the degree of consistency between the microseismic energy feature and the differential feature within that time and space range. A small consistency loss value indicates that the current state conforms to the normal microseismic activity pattern; a large consistency loss value indicates that the current state deviates from the normal pattern and may have an energy accumulation tendency.

[0110] Next, the consistency loss value set is analyzed to determine whether an anomaly exists and whether to trigger an early warning instruction. First, each loss value in the consistency loss value set is compared with a preset consistency loss value threshold to identify all abnormal loss values ​​greater than or equal to the threshold. Then, the monitoring location and time information corresponding to these abnormal loss values ​​are analyzed to determine the spatial distribution characteristics and temporal development trends of the anomaly. When an abnormal loss value is detected and its spatial concentration and duration meet the preset early warning conditions, it is determined that the current microseismic activity has a significant energy accumulation tendency, and the microseismic energy accumulation early warning instruction is immediately triggered. The early warning instruction contains key information such as the time, location, and severity of the anomaly detection. The early warning information is sent to coal mine safety management personnel through various means such as sound and light alarms, text message notifications, and system interface prompts, so that appropriate safety protection and emergency response measures can be taken in a timely manner.

[0111] Furthermore, the energy accumulation tendency analyzer is used to perform consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set to determine a consistency loss value set, including:

[0112] S641: Acquire monitoring coordinates corresponding to consistency loss values ​​greater than or equal to a preset consistency loss value threshold in the consistency loss value set to obtain an abnormal monitoring coordinate set;

[0113] S642: Divide the abnormal monitoring coordinate set into neighboring regions according to a preset monitoring coordinate neighbor threshold to obtain K divided neighboring regions, where K is a positive integer;

[0114] S643: traverse and count the number of abnormal monitoring coordinates in the K divided neighboring areas, and place the statistical results in the areas of the K divided neighboring areas to obtain abnormal coefficients of the K divided neighboring areas;

[0115] S644. When any one of the K divided neighboring region anomaly coefficients is greater than or equal to a preset divided neighboring region anomaly coefficient threshold, a microseismic energy accumulation warning instruction is triggered.

[0116] In a preferred embodiment, first, the spatial location information of the abnormality is identified from the consistency loss value set. Specifically, all loss values ​​in the consistency loss value set are first obtained, and each loss value is compared one by one with the preset consistency loss value threshold. The preset consistency loss value threshold is a critical value determined based on the statistical analysis results of historical normal microseismic activity data. When the consistency loss value calculated in real time exceeds the threshold, it indicates that the consistency between the microseismic energy characteristics and the differential characteristics of the current spatiotemporal position is destroyed, and there is an abnormal energy accumulation tendency. For each abnormal loss value greater than or equal to the preset consistency loss value threshold, the corresponding monitoring coordinate information is extracted, including the three-dimensional spatial coordinates of the location of the microseismic event associated with the loss value. These monitoring coordinates identify the specific spatial location where abnormal microseismic activity occurs in the target coal mine area. By traversing the entire consistency loss value set, the monitoring coordinates corresponding to all abnormal loss values ​​are collected to form an abnormal monitoring coordinate set. This abnormal monitoring coordinate set provides basic data for subsequent spatial aggregation analysis, so that the distribution characteristics and concentration degree of the anomaly can be judged from the spatial dimension.

[0117] Next, a spatial clustering analysis is performed on the abnormal monitoring coordinate set based on a preset monitoring coordinate neighborhood threshold. The goal is to group abnormal monitoring coordinates with similar spatial locations into the same region, laying the foundation for subsequent cumulative risk analysis. The preset monitoring coordinate neighborhood threshold is the critical spatial distance used to determine whether two monitoring coordinates are neighbors. This threshold is set based on the geological structure characteristics of the target coal mine area and the density of microseismic monitoring sensors, with a typical value of 50 to 200 meters. A distance-based clustering algorithm is used to partition the abnormal monitoring coordinate set into neighboring regions. Specifically, the three-dimensional spatial distance between any two coordinate points in the abnormal monitoring coordinate set is calculated. When the distance between the two coordinate points is less than or equal to the preset monitoring coordinate neighborhood threshold, the two coordinate points are considered to belong to the same neighboring region. Through this distance-constrained clustering method, all abnormal monitoring coordinates are grouped according to spatial proximity, with each group forming a partitioned neighboring region. After the neighboring region partitioning process, K partitioned neighboring regions are obtained, where K is a positive integer whose specific value depends on the spatial distribution characteristics of the abnormal monitoring coordinates and the preset neighboring threshold. Each divided neighborhood area contains several abnormal monitoring coordinates with similar spatial positions. This area division method can effectively identify the spatial aggregation pattern of abnormalities and provide an important basis for judging the regional characteristics of energy accumulation.

[0118] Subsequently, the concentration of anomalies in each partitioned neighborhood region is calculated through statistical analysis to obtain a quantitative regional anomaly coefficient. First, the K partitioned neighborhood regions are traversed, and the number of anomaly monitoring coordinates contained in each region is counted. This number reflects the density of anomaly events in that region. Next, the coverage area of ​​each partitioned neighborhood region is calculated—that is, the area of ​​the spatial range formed by all anomaly monitoring coordinates in that region. Next, the number of anomaly monitoring coordinates in each partitioned neighborhood region is divided by the coverage area of ​​that region to obtain the partitioned neighborhood region anomaly coefficient for that region. The partitioned neighborhood region anomaly coefficient is a dimensionless number that reflects the density of anomalies per unit area. Its value directly reflects the concentration of anomalies in that region. A high partitioned neighborhood region anomaly coefficient for a particular region indicates a high number of anomalies per unit area, significant spatial clustering, and a potential indicator of a strong energy accumulation tendency in that region. By performing the same anomaly coefficient calculation on each of the K partitioned neighborhood regions, the K partitioned neighborhood region anomaly coefficients are obtained, providing a quantitative basis for the final early warning judgment.

[0119] Afterward, a final early warning decision is made based on the regional anomaly coefficient, determining whether a microseismic energy accumulation warning instruction needs to be triggered. Specifically, the anomaly coefficients of each of the K partitioned neighboring regions are compared with a preset partitioned neighboring region anomaly coefficient threshold. This threshold is a critical value determined based on statistical analysis of historical microseismic disasters and expert experience. When the regional anomaly coefficient exceeds this threshold, it indicates that the concentration of anomalies within the region has reached a dangerous level that may trigger an energy accumulation risk. If the anomaly coefficient of any of the K partitioned neighboring regions is found to be greater than or equal to the preset partitioned neighboring region anomaly coefficient threshold, a microseismic energy accumulation warning instruction is immediately triggered. This judgment mechanism is based on the concept of regional risk assessment. Even if a single local area has a high-density anomaly concentration, it can pose a significant overall safety threat, necessitating a timely warning. The triggering of the warning instruction includes key information such as the specific location of the anomaly region, the anomaly coefficient value, and the warning level. This provides coal mine safety managers with accurate risk location and severity assessment, facilitating the implementation of targeted safety measures.

[0120] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0121] 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.

[0122] 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 computer, 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.

[0123] 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.

[0124] 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 A step that specifies a function in one or more boxes.

[0125] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0126] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An early warning method based on the tendency of microseismic energy accumulation, characterized in that: The method comprises: Obtain a set of historical microseismic logs collected by an earthquake monitoring sensor array deployed in a target coal mine area within a historical time window, wherein each historical microseismic log includes a log timestamp; Combined with the log timestamps, the historical microseismic log set is processed in stages to obtain a historical microseismic energy characteristic subsequence set; Performing first-order difference operations on the historical microseismic energy characteristic subsequence sets respectively to construct a historical differential energy characteristic subsequence set; Using a spatiotemporal guidance module, trend correlation analysis is performed on the historical microseismic energy characteristic subsequence set and the historical differential energy characteristic subsequence set to obtain a historical microseismic energy trend characteristic set and a historical differential energy trend characteristic set; constructing an energy accumulation tendency analyzer based on the historical microseismic energy trend feature set and the historical differential energy trend feature set; A real-time microseismic log sequence set obtained by real-time monitoring of the seismic monitoring sensor array is obtained, and anomaly identification is performed in combination with the energy accumulation tendency analyzer to trigger a microseismic energy accumulation early warning instruction.

2. The early warning method based on microseismic energy accumulation tendency according to claim 1, characterized in that: Combined with the log timestamps, the historical microseismic log set is processed in stages to obtain a set of historical microseismic energy characteristic subsequences, including: Obtaining a pre-built feature extractor to perform energy feature extraction on the historical microseismic log set to obtain a historical microseismic energy feature set; Serializing the historical microseismic energy feature set according to log timestamps to obtain a historical microseismic energy feature sequence; The historical microseismic energy characteristic sequence is sliced ​​according to a preset sliding window to determine a set of historical microseismic energy characteristic subsequences.

3. The early warning method based on microseismic energy accumulation tendency according to claim 2, characterized in that: Slicing the historical microseismic energy characteristic sequence according to a preset sliding window to determine a set of historical microseismic energy characteristic subsequences includes: Slicing the historical microseismic energy characteristic sequence according to a preset sliding window to obtain an initial historical microseismic energy characteristic subsequence set; A neighboring fusion analysis is performed on the initial historical microseismic energy characteristic subsequence set to obtain the historical microseismic energy characteristic subsequence set.

4. The early warning method based on microseismic energy accumulation tendency according to claim 1, characterized in that: The time-space guidance module is used to perform trend correlation analysis on the historical microseismic energy characteristic subsequence set and the historical differential energy characteristic subsequence set, respectively, to obtain a historical microseismic energy trend characteristic set and a historical differential energy trend characteristic set, including: Based on the preset time scale and sample time-oriented training set, the framework built on the feedforward neural network is supervised trained to obtain the time-oriented submodule; Based on the preset spatial scale and sample spatial orientation training set, the framework built based on the feedforward neural network is supervised and trained to obtain the spatial orientation submodule; The time-oriented submodule and the space-oriented submodule are connected in parallel to obtain the time-space-oriented module.

5. The early warning method based on microseismic energy accumulation tendency according to claim 4, characterized in that: include: The time-oriented submodule and the space-oriented submodule are used to extract spatiotemporal features of the historical microseismic energy feature sequence and the historical differential energy feature sequence, respectively, to obtain historical microseismic energy time trend features, historical microseismic energy space trend features, historical differential energy time trend features, and historical differential energy space trend features; Performing trend correlation analysis on the historical microseismic energy time trend characteristics and the historical microseismic energy spatial trend characteristics to obtain historical microseismic energy trend characteristics; The trend correlation analysis is performed on the historical differential energy time trend characteristics and the historical differential energy space trend characteristics to obtain the historical differential energy trend characteristics.

6. The early warning method based on microseismic energy accumulation tendency according to claim 5, characterized in that: Performing trend correlation analysis on the historical microseismic energy temporal trend characteristics and the historical microseismic energy spatial trend characteristics to obtain historical microseismic energy trend characteristics, including: Calculating feature similarities of the historical microseismic energy time trend feature and the historical microseismic energy spatial trend feature to obtain a feature similarity set; Normalizing the feature similarity set and constructing the processing results into a matrix to obtain a historical microseismic energy-time adjacency matrix; The historical microseismic energy time adjacency matrix and the historical microseismic energy spatial trend feature are convolved to obtain the historical microseismic energy trend feature.

7. The early warning method based on microseismic energy accumulation tendency according to claim 1, characterized in that: An energy accumulation tendency analyzer is constructed based on the historical microseismic energy trend feature set and the historical differential energy trend feature set, including: Taking each historical microseismic energy trend feature and each historical differential energy trend feature in the historical microseismic energy trend feature set and the historical differential energy trend feature set as a set of positive sample pairs, to obtain a positive sample pair set; Mapping the historical microseismic energy trend feature set and the historical differential energy trend feature set into a historical microseismic energy trend feature probability distribution and a historical differential energy trend feature probability distribution, respectively; Performing symmetric KL divergence calculation on the historical microseismic energy trend characteristic probability distribution and the historical differential energy trend characteristic probability distribution to determine a consistency loss function; By taking the positive sample pair set as contrast representation learning input, a framework based on a convolutional neural network is supervised and calculated, and an energy accumulation tendency analyzer is constructed by minimizing a consistency loss function.

8. The early warning method based on microseismic energy accumulation tendency according to claim 7, characterized in that: Mapping the historical microseismic energy trend feature set and the historical differential energy trend feature set into a historical microseismic energy trend feature probability distribution and a historical differential energy trend feature probability distribution, respectively, includes: Normalizing and softmax mapping the historical microseismic energy trend feature set to obtain a probability distribution of the historical microseismic energy trend features; The historical differential energy trend feature set is normalized and softmax mapped to obtain a historical differential energy trend feature probability distribution.

9. The early warning method based on microseismic energy accumulation tendency according to claim 1, characterized in that: Acquiring a real-time microseismic log sequence set obtained by real-time monitoring of the seismic monitoring sensor array, performing anomaly identification in combination with the energy accumulation tendency analyzer, and triggering a microseismic energy accumulation early warning instruction, including: performing energy feature extraction on the real-time microseismic log sequence set to determine a real-time microseismic energy feature sequence set; performing a first-order difference operation on the real-time microseismic energy characteristic sequence set to determine a real-time differential energy characteristic sequence set; Performing trend correlation analysis on the real-time microseismic energy feature sequence set and the real-time differential energy feature sequence set using a time-space guidance module to determine a real-time microseismic energy trend feature set and a real-time differential energy trend feature set; Performing consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set using the energy accumulation tendency analyzer to determine a consistency loss value set; The consistency loss value set is analyzed, and if an anomaly is found, a microseismic energy accumulation warning instruction is triggered.

10. The early warning method based on microseismic energy accumulation tendency according to claim 9, characterized in that: Performing consistency loss value analysis on the real-time microseismic energy trend feature set and the real-time differential energy trend feature set using the energy accumulation tendency analyzer to determine a consistency loss value set includes: Acquire monitoring coordinates corresponding to consistency loss values ​​greater than or equal to a preset consistency loss value threshold in the consistency loss value set to obtain an abnormal monitoring coordinate set; Divide the abnormal monitoring coordinate set into neighboring regions according to a preset monitoring coordinate neighbor threshold to obtain K divided neighboring regions, where K is a positive integer; Traversing and counting the number of abnormal monitoring coordinates in the K divided neighboring areas, and placing the statistical results in the areas of the K divided neighboring areas to obtain abnormal coefficients of the K divided neighboring areas; When any one of the K divided neighboring region anomaly coefficients is greater than or equal to a preset divided neighboring region anomaly coefficient threshold, a microseismic energy accumulation early warning instruction is triggered.

Images