Mine deep well ground stress real-time monitoring system and method

Abstract

The invention discloses a mine deep well crustal stress real-time monitoring system and method, and relates to the technical field of mine deep well crustal stress monitoring, and the method specifically comprises the following steps: collecting crustal stress monitoring point data in each sub-block, and generating a regional stress distribution map based on the spatial relationship between monitoring points; based on the generated regional stress distribution map, evaluating whether the division mechanism of the current risk identification region adapts to the current crustal stress distribution state; dividing each sub-block into a first identification matching area, a second identification offset area and a third identification reconstruction area according to the evaluation result, and dynamically adjusting the division boundary of the risk identification area according to the classification result; and executing corresponding risk response measures based on the divided region types, and dynamically correcting the risk identification range based on a response result. According to the method, the problem of ground stress identification lag is solved, and the intelligent monitoring effect of risk area dynamic division and accurate regulation and control is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of in-situ stress monitoring in deep mines, and particularly to a real-time monitoring system and method for in-situ stress in deep mines. Background Art

[0002] In-situ stress in deep mines refers to the stress state inside rock masses caused by factors such as geological structures, rock formation burial depth, and mining activities in the deep mine environment. It is directly related to the stability of roadway surrounding rocks, the safe operation of equipment, and the risk control of personnel operations. As the mining depth of mines continuously increases, the in-situ stress shows a non-linear increasing characteristic, which is extremely likely to induce major geological disasters such as rock bursts, roof collapses, and rock bursts, seriously threatening the safety of mine production and the lives and property of personnel. Therefore, carrying out real-time monitoring of in-situ stress in deep mines can achieve continuous perception and dynamic early warning of stress changes, help to timely detect stress anomalies, predict potential risks, and take targeted prevention and control measures. Compared with traditional fixed-point sampling or manual inspection methods, real-time monitoring not only greatly improves the timeliness and accuracy of data acquisition, but also can continuously track and trend analyze the in-situ stress state through remote communication and intelligent algorithms, providing key data support and decision-making basis for intelligent management and disaster prevention and control in mines.

[0003] The existing real-time monitoring technology for in-situ stress in deep mines mainly relies on a multi-point distributed sensing system. By deploying high-sensitivity in-situ stress sensors (such as strain gauges, fiber optic sensors, acoustic emission sensors, etc.) at key positions underground, stress change data inside the rock mass are collected in real time; these data are preliminarily processed and cached by a local data acquisition module, and then transmitted to the ground monitoring center through industrial wireless networks, fiber optic communication, or mine-specific wired networks; the ground system calibrates, filters, models, and analyzes the original data through an embedded platform or a host computer system, and conducts trend prediction and anomaly identification in combination with historical data and geological models; finally, the real-time in-situ stress change curve, anomaly alarm information, spatial distribution map, etc. are presented to mine management personnel through a visualization platform to achieve dynamic supervision of in-situ stress changes. The whole process includes five core links: stress acquisition, signal conversion, data transmission, intelligent analysis, and result display, forming a complete end-to-end, highly continuous, and highly automated real-time monitoring technology system for in-situ stress.

[0004] The existing technology has the following deficiencies:

[0005] In the case of continuous advancement of mine tunneling operations, the rock mass in the front gradually accumulates stress due to continuous disturbance, while the rear area undergoes unloading due to stress release, resulting in an obvious directional imbalance in stress distribution before and after the tunneling face. This imbalance often manifests as significant stress concentration in local areas, but the average stress value of the entire area changes insignificantly, which is easily masked by traditional identification methods. In this situation, the existing real-time monitoring technology for in-situ stress in deep mines cannot dynamically adjust the risk identification scope division mechanism according to the degree of stress distribution imbalance during tunneling. Most of its risk assessments rely on the overall stress mean or statistical fluctuation range, and it is unable to identify the stress evolution trend with spatial differences and direction concentration, resulting in a deviation between the boundary of the risk area delimited by the system and the actual stress concentration zone, further missing high-risk sections, being unable to trigger effective early warning or support responses, and posing serious safety hazards.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure. Therefore, it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The object of the present invention is to provide a real-time monitoring system and method for in-situ stress in deep mines to solve the problems in the above background art.

[0008] To achieve the above object, the present invention provides the following technical solution: A real-time monitoring method for in-situ stress in deep mines, specifically including the following steps:

[0009] Construct a spatial sub-block model of the tunneling area, dynamically divide the tunneling area based on the advancing position of the tunneling face and the distribution information of monitoring points, and establish a spatial identifier for each divided sub-block.

[0010] Collect the in-situ stress monitoring point data within each sub-block, and generate a regional stress distribution map based on the spatial relationship between the monitoring points.

[0011] Based on the generated regional stress distribution map, evaluate whether the current risk identification area division mechanism is suitable for the current in-situ stress distribution state.

[0012] According to the evaluation results, divide each sub-block into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area, and dynamically adjust the division boundary of the risk identification area according to the classification results.

[0013] Based on the divided regional types, execute corresponding risk response measures, and dynamically correct the risk identification scope based on the response results.

[0014] Store the monitoring data, identification results, and response behaviors during the division adjustment process, and continuously optimize the division mechanism of the risk identification area based on the stored historical data.

[0015] Preferably, collect the data of the in-situ stress monitoring points in each sub-block, and generate a regional stress distribution map based on the spatial relationship between the monitoring points. Specifically: collect the real-time stress monitoring values and their three-dimensional spatial coordinate information of all the in-situ stress monitoring points located in each sub-block, calculate the stress difference and spatial distance between adjacent monitoring points, construct a polygon topological structure according to the spatial relationship of each monitoring point, and perform interpolation fitting processing on the monitoring point data within this topological structure, and further perform grid operation to generate a continuous regional stress distribution map for characterizing the trend of the in-situ stress gradient change between different sub-blocks in the tunneling direction.

[0016] Preferably, based on the generated regional stress distribution map, evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state, which specifically includes the following steps:

[0017] Based on the generated regional stress distribution map, extract the stress structure feature information of each sub-block from it and perform preprocessing on it;

[0018] Extract the directional stress distribution information and stress spatial discretization information from the preprocessed stress structure feature information of each sub-block, and analyze them to generate the directional stress gradient index and stress discretization coefficient of each sub-block respectively;

[0019] Construct a weighted model for the generated directional stress gradient index and stress discretization coefficient of each sub-block, generate the division adaptation evaluation coefficient of each sub-block through weighted summation, and generate the comprehensive evaluation coefficient through the standard deviation calculation formula;

[0020] Determine the preset comprehensive evaluation coefficient threshold, and compare it with the generated comprehensive evaluation coefficient after determination, and evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state according to the comparison result.

[0021] Preferably, the acquisition logic of the directional stress gradient index of each sub-block is as follows:

[0022] Extract the directional stress distribution information from the preprocessed stress structure feature information of each sub-block, specifically including the average stress value of the front-edge monitoring point group, the average stress value of the rear-edge monitoring point group, and the length in the tunneling direction of each sub-block in the generated regional stress distribution map, and label them as QJP i 、HJP i and L i ,QJP iDenotes the average stress value of the front monitoring point group of the i-th sub-block in the generated regional stress distribution map, HJP i Denotes the average stress value of the rear monitoring point group of the i-th sub-block in the generated regional stress distribution map, L i Denotes the driving direction length of the i-th sub-block in the generated regional stress distribution map, i = 1, 2, 3, …, k, where k is a positive integer;

[0023] Calculate the directional stress gradient index of each sub-block. The specific calculation formula is as follows:

[0024]

[0025] In the formula, DSGI i Is the directional stress gradient index of the i-th sub-block, QJP i Denotes the average value of the stress values of all monitoring points at the front boundary in the driving direction of the i-th sub-block in the generated regional stress distribution map, HJP i Denotes the average value of the stress values of all monitoring points at the rear boundary of the i-th sub-block in the generated regional stress distribution map, L i Denotes the physical length of the i-th sub-block in the driving direction in the generated regional stress distribution map.

[0026] Preferably, the acquisition logic of the stress dispersion coefficient of each sub-block is as follows:

[0027] Extract the stress spatial dispersion information from the stress structure characteristic information of each preprocessed sub-block, specifically including the stress value of each monitoring point in each sub-block in the generated regional stress distribution map and the distance from each monitoring point to the geometric center of the sub-block to which it belongs, and mark them respectively as And Denotes the stress value of the j-th monitoring point in the i-th sub-block in the generated regional stress distribution map, Denotes the distance from the j-th monitoring point in the i-th sub-block in the generated regional stress distribution map to the geometric center of the sub-block to which it belongs, i = 1, 2, 3, …, k, j = 1, 2, 3, …, h, where both k and h are positive integers;

[0028] Calculate the stress dispersion coefficient of each sub-block. The specific calculation formula is as follows:

[0029]

[0030] In the formula, SDC i Is the stress dispersion coefficient of the i-th sub-block.

[0031] Preferably, for the directional stress gradient index DSGI of each generated sub-block i And the stress dispersion coefficient SDCi Construct a weighted model and generate the division adaptation evaluation coefficient of each sub-block through weighted summation. The specific calculation formula is as follows:

[0032] PAEC i = ω1 * DSGI i + ω2 * SDC i

[0033] In the formula, PAEC i is the division adaptation evaluation coefficient of the i-th sub-block, and ω1 and ω2 are the non-zero weight coefficients of the directional stress gradient index DSGI i and the stress dispersion coefficient SDC i respectively, and ω1 + ω2 = 1;

[0034] Perform standard deviation analysis on the division adaptation evaluation coefficient PAEC i of each sub-block to generate a comprehensive evaluation coefficient. According to the formula:

[0035]

[0036] In the formula, CEC is the comprehensive evaluation coefficient, i = 1, 2, 3,..., k, and k is a positive integer.

[0037] Preferably, determine a preset comprehensive evaluation coefficient threshold CEC yuzhi , and compare it with the generated comprehensive evaluation coefficient CEC after determination. Evaluate whether the division mechanism of the current risk identification area adapts to the current in-situ stress distribution state according to the comparison result. The specific comparison analysis is as follows:

[0038] If CEC ≤ CEC yuzhi , the division mechanism of the current risk identification area adapts to the current in-situ stress distribution state;

[0039] If CEC > CEC yuzhi , the division mechanism of the current risk identification area does not adapt to the current in-situ stress distribution state.

[0040] Preferably, if the evaluation result is that the division mechanism of the current risk identification area does not adapt to the current in-situ stress distribution state, determine a preset division adaptation evaluation coefficient threshold interval [PAEC min , PAEC max , and compare it with the generated division adaptation evaluation coefficient PAEC i of each sub-block after determination. Divide each sub-block into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area according to the comparison result. The specific division is as follows:

[0041] If PAEC i < PAECmin , divide the sub-block into a first recognition and matching area;

[0042] If PAEC min ≤PAEC i ≤PAEC max , divide the sub-block into a second recognition offset area;

[0043] If PAEC i >PAEC max , divide the sub-block into a third recognition reconstruction area;

[0044] Dynamically adjust the division boundary of the risk recognition area according to the classification results. Specifically: for the sub-blocks divided into the first recognition and matching area, keep their original risk recognition boundaries unchanged and continue to use the current recognition mechanism for monitoring and evaluation in subsequent cycles; for the sub-blocks divided into the second recognition offset area, set a buffer expansion area on their current boundaries, expand their recognition boundary ranges, and increase the sensitivity parameters of the recognition mechanism; for the sub-blocks divided into the third recognition reconstruction area, reconstruct their original recognition boundaries, reset the risk judgment parameters, and enable a dynamic recognition mechanism for the sub-blocks to correct their division boundaries in real time.

[0045] Preferably, based on the divided area types, execute corresponding risk response measures and dynamically correct the risk recognition scope based on the response results. Specifically:

[0046] For the sub-blocks divided into the first recognition and matching area, maintain the current risk response strategy unchanged, adopt the conventional monitoring frequency and the original support strategy, and record the monitoring data in subsequent cycles as the reference for evaluation;

[0047] For the sub-blocks divided into the second recognition offset area, execute an enhanced response strategy, including increasing the monitoring frequency, introducing auxiliary sensors and early warning algorithms, obtaining dynamic feedback data, and adjusting the boundary shape and coverage ratio of the risk recognition scope when there is a deviation between the recognition result and the actual response;

[0048] For the sub-blocks divided into the third recognition reconstruction area, execute key early warning and active intervention measures, including real-time monitoring data access, improvement of the support strength level, and reset of the recognition model parameters, and continuously compare the risk response effect with the stress evolution trend in subsequent cycles, and dynamically correct the risk recognition boundary of the sub-block periodically according to the offset result.

[0049] Preferably, a real-time in-situ stress monitoring system for deep mines includes a spatial division and modeling module, a stress map generation module, a recognition and adaptation evaluation module, a block type discrimination module, a partition response correction module, and a recognition mechanism optimization module;

[0050] The spatial division modeling module constructs a spatial sub-block model of the tunneling area, dynamically divides the tunneling area based on the advancing position of the tunneling face and the distribution information of monitoring points, and establishes a spatial identifier for each divided sub-block;

[0051] The stress map generation module collects the data of in-situ stress monitoring points in each sub-block and generates a regional stress distribution map based on the spatial relationship between the monitoring points;

[0052] The identification adaptation evaluation module evaluates whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state based on the generated regional stress distribution map;

[0053] The block type discrimination module divides each sub-block into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area according to the evaluation results, and dynamically adjusts the division boundary of the risk identification area according to the classification results;

[0054] The partition response correction module executes corresponding risk response measures based on the divided area types, and dynamically corrects the risk identification scope based on the response results;

[0055] The identification mechanism optimization module stores the monitoring data, identification results, and response behaviors during the division adjustment process, and continuously optimizes the division mechanism of the risk identification area based on the stored historical data.

[0056] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0057] 1. By constructing a spatial sub-block model of the tunneling area and combining the distribution data of monitoring points with the tunneling advancement position, the present invention first realizes the spatial fine division and dynamic modeling of the stress state in the tunneling front area, significantly improving the system's structural perception ability of local stress changes. By introducing two key indicators, namely the directional stress gradient index and the stress dispersion coefficient, and combining with a weighted model to generate a division adaptation evaluation coefficient, and then generating a comprehensive evaluation coefficient through standard deviation analysis, the quantification evaluation of the uneven degree of in-situ stress distribution is realized, effectively overcoming the "recognition blind area" problem of the traditional average value judgment method for local high-stress aggregation paragraphs. Thus, the present invention significantly enhances the response ability of the risk identification mechanism to the directional stress evolution trend, ensures that high-risk sub-blocks can be accurately identified, and provides precise identification boundary support for subsequent response regulation.

[0058] 2. Based on the recognition mechanism, the present invention further constructs a regional division strategy centered on the "first recognition and matching area, second recognition offset area, and third recognition reconstruction area", and designs differentiated risk boundary adjustment and response control strategies for different types of regions. The system can perform fine classification based on the evaluation coefficients of each sub-block and the preset threshold range, and automatically execute operations such as matching maintenance, buffer expansion, or recognition reconstruction on various regions, realizing the dynamic correction and real-time reconstruction capabilities of the risk recognition boundary. This strategy not only significantly improves the response accuracy to the high stress concentration area of tunneling disturbance, but also effectively reduces the probability of false alarms and missed alarms, enabling the risk recognition mechanism to have the intelligent adaptation ability synchronized with geological evolution, and is applicable to continuous tunneling operations in complex in-situ stress environments.

[0059] 3. The present invention also designs a full-process storage and historical backtracking mechanism for the data in the risk recognition process, structurally records the monitoring data, recognition results, and response behaviors in each cycle, and continuously mines their evolution laws through a software system for dynamically optimizing the judgment parameters and boundary adjustment strategies of the recognition model in the next cycle. This mechanism realizes the transformation from a "static strategy execution" to a "feedback-driven self-evolution" intelligent recognition mechanism, has the ability to continuously optimize the recognition strategy based on the historical operation effect, and significantly enhances the long-term reliability and engineering adaptability of the in-situ stress monitoring system. Through data accumulation and model fine-tuning, the present invention can construct an in-situ stress recognition platform with domain memory ability, ensuring that the system always has high recognition accuracy and high response efficiency under variable geological backgrounds and continuous disturbance conditions, and truly realizing the technological leap of mine deep well in-situ stress monitoring from "passive recognition" to "intelligent control". BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0061] Figure 1 It is a schematic flow chart of a real-time in-situ stress monitoring system and method for mine deep wells of the present invention.

[0062] Figure 2 It is a schematic module diagram of a real-time in-situ stress monitoring system and method for mine deep wells of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0064] The present invention provides a method for real-time monitoring of in-situ stress in deep mines as shown in Figure 1 and specifically includes the following steps:

[0065] Construct a spatial sub-block model of the tunneling area, dynamically divide the tunneling area based on the advancing position of the tunneling face and the distribution information of monitoring points, and establish a spatial identifier for each divided sub-block;

[0066] The construction of the spatial sub-block model of the tunneling area can be achieved by grid modeling of the three-dimensional spatial information of the tunneling roadway. Specifically, first, obtain its spatial geometric shape information based on the initial design drawings or BIM models of the roadway, and establish a local coordinate system with the tunneling face as the reference point; then, according to the set spatial step size or unit volume parameter, perform regular or adaptive partitioning of this space along the tunneling direction and the lateral sides on both sides, generating a set of sub-block sets with spatial boundary attributes. Each sub-block, as a spatial entity, includes information such as the spatial three-dimensional coordinate range, relative position in the entire tunneling area, adjacency relationship, etc.; this operation can be implemented through a spatial grid division algorithm, a three-dimensional modeling library, or a geological modeling engine, and is stored as a structured spatial grid data model by software for subsequent in-situ stress data binding and dynamic analysis.

[0067] The core of dynamic division lies in enabling the update logic of the spatial sub-blocks to respond in real time to changes in the position of the tunneling face and adjustments in the distribution of monitoring points. The specific implementation method is that the system receives the position advancement data of the tunneling equipment in real time and dynamically associates it with the spatial sub-block model; once the tunneling face advances forward beyond the set threshold distance, the system automatically triggers a re-division mechanism: taking the current tunneling face as the benchmark, reconstruct the sub-blocks within a certain scale range in front of and behind it, and at the same time, according to the three-dimensional coordinates of each current monitoring point, associate it with the nearest spatial sub-block. Subsequently, the system assigns a unique spatial identifier (such as encoding based on the tunneling coordinates + block number rule) to each newly generated or updated sub-block, and establishes a mapping relationship between the sub-block and the monitoring point. These operations are comprehensively executed by the spatial management module in combination with the tunneling time series data, sensor distribution data, and sub-block layer data, and are completed at the software level for dynamic identification and update.

[0068] By constructing a spatial sub-block model and combining tunneling propulsion with the distribution of monitoring points for dynamic division, it is possible to achieve structured management and refined positioning of in-situ stress data in the tunneling area. Traditional in-situ stress monitoring systems often rely on static area division and cannot adapt to the monitoring mismatch problems caused by the continuous change of the spatial structure during tunneling, which easily leads to the disconnection between data distribution and the actual risk area. Through this spatial sub-block mechanism, the system can synchronize the corresponding relationship between the geographical space and the data space in real time, so that each monitoring data is accurately located within a dynamic block, facilitating subsequent construction of stress distribution maps, regional assessment, risk level judgment, and execution of the dynamic division mechanism in units of "sub-blocks". At the same time, the establishment of spatial identifiers also provides unified data structure support for the comparative analysis of historical data, sustainable training of identification models, and subsequent multiple rounds of dynamic division, enabling the system to have the capabilities of spatial continuity, high resolution, and time-effect adaptability, thus more effectively coping with the problem of uneven stress distribution in the tunneling environment.

[0069] Collect the in-situ stress monitoring point data within each sub-block and generate a regional stress distribution map based on the spatial relationship between the monitoring points;

[0070] In this embodiment, collecting the in-situ stress monitoring point data within each sub-block and generating a regional stress distribution map based on the spatial relationship between the monitoring points specifically means: collecting the real-time stress measurement values and their three-dimensional spatial coordinate information of all in-situ stress monitoring points located within each sub-block, calculating the stress difference and spatial distance between adjacent monitoring points, constructing a polygon topological structure according to the spatial relationship of each monitoring point, and performing interpolation fitting processing on the monitoring point data within this topological structure, and further performing grid operation to generate a continuous regional stress distribution map for characterizing the in-situ stress gradient change trend between different sub-blocks in the tunneling direction.

[0071] In practical applications, multiple in-situ stress monitoring points are usually arranged in each sub-block of the tunneling area. These monitoring points are connected to the ground data acquisition system by wired or wireless means and continuously report their real-time stress measurement values and preset coordinate positions. After the system receives the stress data of each monitoring point, it extracts the current stress value of each point through data decoding and parsing functions and associates it with the pre-stored three-dimensional coordinate information. Subsequently, the system calls the built-in spatial analysis algorithm, calculates the Euclidean distance between each point and its adjacent point in the tunneling direction based on the coordinate relationship between the monitoring points, and synchronously calculates the stress difference to obtain a set of "point-to-point" physical state change sequences. This processing method not only constructs the spatial-physical structure relationship of the original data but also provides a high-resolution data basis for subsequent regional map modeling, stress gradient assessment, and uneven state identification. The core purpose of adopting this method is to achieve dynamic association of spatial data and quantitative expression of stress differences at the software level, avoid misjudgment caused by single-point value distortion, and truly restore the spatial characteristics of stress evolution on the tunneling face.

[0072] After completing the data collection and difference calculation of the monitoring points, the system constructs a polygon topology structure based on the spatial coordinates of each monitoring point. This structure can be automatically generated in the software through the Delaunay triangulation algorithm or the Voronoi partitioning algorithm to ensure the formation of a topologically closed adjacent relationship graph among all monitoring points. The role of the polygon topology structure is to organize the discrete monitoring points into a spatial grid with geometric adjacency relationships, enabling each point to participate in the fitting of local spatial functions through its adjacent point relationships. Subsequently, the system performs interpolation fitting processing on the stress values of the monitoring points in this structure, often using mathematical methods such as inverse distance weighting (IDW), Kriging interpolation, or spline function interpolation, to numerically predict the unknown area based on the distance and directional relationships of the points. After the fitting is completed, the system will perform a grid operation in the topology structure at a fixed step size, that is, divide the entire space into regular or irregular cells, so that there is a corresponding fitting value in each grid cell, and finally form a continuous regional stress distribution map. The entire process requires no manual intervention and can be efficiently implemented in the software through a data processing engine, a graphics calculation library, or a geographic information system module.

[0073] The method of constructing a topology structure based on the spatial relationships of monitoring points and generating a continuous regional map by combining interpolation and grid operation aims to break through the limitations of the traditional "discrete measurement point interpretation" and achieve visual, continuous, and directional expression of in-situ stress distribution. During the tunneling process, the stress change is not an isolated point change but a dynamic process of diffusion along the spatial direction. Single-point judgment will lead to the omission of local high-stress concentrations. Through the above modeling process, the system can fuse the data of multiple measurement points into a continuous spatial field, reflect the real-time change trend of the stress gradient in the area, and clearly display uneven characteristics such as high in the front and low in the back or spatial offset. This not only improves the overall cognitive ability of the risk situation but also provides an intuitive and reliable quantitative basis for subsequent risk area division, boundary dynamic adjustment, and response strategy selection. This mechanism is completed at the software level, does not rely on external devices or manual identification, and has high scalability and automated operation capabilities, which is a key supporting means for realizing refined monitoring and intelligent risk identification.

[0074] Based on the generated regional stress distribution map, evaluate whether the current risk identification area division mechanism is suitable for the current in-situ stress distribution state;

[0075] In this embodiment, based on the generated regional stress distribution map, evaluating whether the current risk identification area division mechanism is suitable for the current in-situ stress distribution state specifically includes the following steps:

[0076] Based on the generated regional stress distribution map, extract the stress structure feature information of each sub-block and preprocess it;

[0077] After generating the regional stress distribution map, the system can perform data extraction and processing at the sub-block level based on the spatial boundary model of the sub-blocks. The specific method is as follows: First, the system performs a spatial slicing operation on the stress map according to the spatial boundary coordinates of the sub-blocks, and cuts the entire map into several map segments according to the sub-block boundaries; then, in each map segment, the system identifies the coordinates of all corresponding monitoring points and stress value information, and constructs a point set data structure. For each sub-block, the system statistically analyzes the local stress change trend in its map segment, and automatically calculates a series of basic physical quantities such as the stress mean value of the monitoring points at the front and rear boundaries of the sub-block, the boundary distance, the stress mean value of all internal monitoring points, and the distance from each monitoring point to the center point. These numerical data constitute the stress structure characteristic information of the sub-block, which can be further used for gradient and discreteness modeling. The entire process can be implemented in the software platform through map indexing technology, spatial mapping functions, and data extraction modules, without manual intervention, and has a high degree of automation.

[0078] The purpose of preprocessing the extracted stress structure characteristic information is to improve the accuracy and robustness of subsequent parameter calculations, and prevent noise, missing measurements, or extreme outliers in the original map data from interfering with the evaluation model. First, in the preprocessing stage, the system checks the integrity of the stress point set within each sub-block, and automatically eliminates data points with overlapping spatial positions, zero or invalid stress values; second, the system applies local anomaly detection algorithms (such as based on IQR or Z-score) to identify stress anomaly points, and uses the average value of adjacent regions or interpolation methods to correct or replace them; third, unit standardization or normalization operations are performed on all stress values to make them comparable among different regions; in addition, the system also unifies the stress gradient calculation direction, confirms the consistency of the tunneling direction axis and the sub-block projection, and ensures that the direction parameters are solved under the same reference system. The entire preprocessing process is jointly executed in the software through a data cleaning module, an outlier correction logic, and a standardization tool chain, as a key data integration step before parameter calculation, ensuring that the evaluation results are stable, accurate, and interpretable.

[0079] Extract the directional stress distribution information and stress spatial discreteness information from the stress structure characteristic information of each preprocessed sub-block, and analyze them to generate the directional stress gradient index and stress discreteness coefficient of each sub-block respectively;

[0080] After obtaining and preprocessing the stress structure feature information of each sub-block, the system can extract the directional stress distribution information and stress space discretization information from it through the built-in spatial analysis and numerical decomposition module. The specific extraction method is as follows: First, the system locates the set of monitoring points at the front and rear edges of the block according to the geometric boundaries of the sub-block in the tunneling direction, calculates the stress means at both ends respectively, and combines the tunneling length of the block. Through the difference operation, the directional stress distribution information of the sub-block is extracted to reflect the stress change trend at the front and rear boundaries; Subsequently, the system performs spatial aggregation processing on all the monitoring points within the sub-block, obtains the physical distances of each monitoring point relative to the geometric center of the block, and combines the corresponding stress values. Through spatial statistical methods, the weighted deviation of the point set is calculated, and the stress space discretization information of the block is extracted to characterize the uniformity and fluctuation characteristics of the internal stress distribution. The extraction process of the above two types of information is completely realized based on the existing structure feature data and spatial position relationship, and the system automatically executes through the spatial operation and data mining tool chain to ensure that the extraction results have quantitative characteristics and scalability, providing accurate data input for the subsequent construction of the judgment model.

[0081] Construct a weighted model for the directional stress gradient index and stress dispersion coefficient of each generated sub-block, generate the division adaptation evaluation coefficient of each sub-block through weighted summation, and generate the comprehensive evaluation coefficient through the standard deviation calculation formula;

[0082] Determine the preset threshold of the comprehensive evaluation coefficient, and compare it with the generated comprehensive evaluation coefficient after determination. According to the comparison result, evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state.

[0083] Determining the preset comprehensive evaluation coefficient threshold can be automatically generated through the historical data modeling module built into the system in combination with multiple rounds of simulation analysis. The specific implementation method is as follows: The system first calls the stress distribution maps of multiple typical regions at different stages in the historical tunneling data and the corresponding risk identification and division effect records, and constructs the corresponding relationship between the comprehensive evaluation coefficients in these sample data and the actual division effects as a training data set; Subsequently, the system conducts clustering analysis and statistical modeling on the comprehensive evaluation coefficients of all historical samples to identify the distribution intervals of the evaluation coefficients in the effective and ineffective states of the identification mechanism; On this basis, the system calculates the boundary values for dividing the "adaptive" and "non - adaptive" states by setting a specific confidence level (such as a 95% confidence interval), and sets this boundary value as the comprehensive evaluation coefficient threshold. At the same time, to enhance dynamic adaptability, the system also supports combining real - time operation data and using a sliding window retraining mechanism to periodically correct the threshold, so as to ensure that this threshold always has representativeness and discriminative power at different tunneling stages and different geological backgrounds. The entire process is jointly completed by the statistical modeling tool, risk mapping module, and data update mechanism in the software, realizing the automatic generation and dynamic optimization of the threshold without manual intervention.

[0084] In this embodiment, the acquisition logic of the directional stress gradient index for each sub - block is as follows:

[0085] Extract the directional stress distribution information from the stress structure characteristic information of each pre - processed sub - block, specifically including the average stress value of the front - edge monitoring point group, the average stress value of the rear - edge monitoring point group, and the tunneling direction length of each sub - block in the generated regional stress distribution map, and label them as QJP i , HJP i and L i , QJP i represents the average stress value of the front - edge monitoring point group of the i - th sub - block in the generated regional stress distribution map, HJP i represents the average stress value of the rear - edge monitoring point group of the i - th sub - block in the generated regional stress distribution map, L i represents the tunneling direction length of the i - th sub - block in the generated regional stress distribution map, i = 1, 2, 3, …, k, where k is a positive integer;

[0086] After generating the regional stress distribution map, the system can automatically extract three types of core quantitative data, namely, the average stress value of the front-edge monitoring point group, the average stress value of the rear-edge monitoring point group, and the length in the tunneling direction of each sub-block, based on the spatial structure of the sub-blocks and the distribution information of the monitoring points, through map parsing and spatial indexing methods. The specific implementation method is as follows: First, the system calls the spatial boundary model of the sub-blocks to identify the "front-edge boundary" and "rear-edge boundary" of each sub-block in the tunneling direction, and then performs spatial matching between the monitoring point coordinates in the map and the sub-block boundary positions to extract all the monitoring point sets located within a specified distance threshold near these two end boundaries. The system calculates the mean value based on the stress values of these monitoring points to obtain the average stress value of the front-edge monitoring point group (used to reflect the stress concentration at the front end of the sub-block) and the average stress value of the rear-edge monitoring point group (used to reflect the stress release or unloading trend at the rear end); while the length in the tunneling direction is automatically calculated by the system according to the starting and ending coordinates of the sub-block in the main tunneling axis direction, representing the actual physical length between the "front-rear" boundaries of the block in space, and is used to quantify the distance scale of stress change. The extraction and calculation processes of these data can be batch completed in the software platform through algorithms such as spatial geometry processing, coordinate screening, and stress value aggregation, featuring high scalability and automation, and are an indispensable quantitative basis for subsequent calculation of the directional stress gradient index.

[0087] Calculate the directional stress gradient index of each sub-block. The specific calculation formula is as follows:

[0088]

[0089] In the formula, DSGI i is the directional stress gradient index of the i-th sub-block, QJP i represents the average value of the stress values of all the monitoring points at the front boundary in the tunneling direction of the i-th sub-block in the generated regional stress distribution map, HJP i represents the average value of the stress values of all the monitoring points at the rear boundary of the i-th sub-block in the generated regional stress distribution map, L i represents the physical length of the i-th sub-block in the tunneling direction in the generated regional stress distribution map.

[0090] The calculation of this directional stress gradient index adopts the method of normalizing the stress difference between the front edge and the rear edge of the sub-block in the tunneling direction and performing non-linear mapping through a logarithmic function. Its essential purpose is to accurately depict the directional intensity and gradient steepness of stress change of each sub-block in the tunneling direction. First, |QJP i -HJP i | 2represents the square difference between the average stress values of the front and rear edges, which is intended to emphasize the absolute size of the stress difference. At the same time, the square operation is used to enhance the significance of large differences and compress the interference of small differences, making local strong gradients easier to identify. Secondly, the square difference is divided by (L i ) 2 The purpose is to normalize the stress difference to the gradient intensity per unit length, so that the parameter is not directly affected by the size of the sub-block and achieve comparability; secondly, adding 1 to the fractional structure and taking the natural logarithm is a nonlinear compression method commonly used in geological stress analysis. It can prevent numerical overflow caused by extreme gradients and improve the resolution of small and medium gradient sections, so that the overall result has better mathematical continuity and engineering interpretability. Through such a calculation structure, the DSGI generated in the end i The value can comprehensively reflect the stress imbalance of the sub-block in the excavation direction, reflecting both the amplitude of the change and the spatial gradient characteristics. It is an important basis for the subsequent adaptability assessment and risk division of the sub-block.

[0091] Directional stress gradient index DSGI of the i-th sub-block i The value of DSGI directly reflects the degree of imbalance of local stress in the tunneling direction of the sub-block, thus playing a core supporting role in evaluating whether the current risk identification area division mechanism is suitable for the current ground stress distribution state. i The larger the value, the more dramatic the stress change between the front and rear boundaries of the sub-block, showing obvious stress concentration or unloading phenomenon, indicating that there may be a high-risk stress disturbance zone in the area that has not been effectively identified by the current division mechanism; if the DSGI of multiple adjacent sub-blocks i If the value fluctuates violently or deviates significantly from the average level, it usually means that the overall division mechanism fails to fully capture the directional evolution trend of stress, and there is a deviation between the defined risk identification boundary and the actual stress gradient distribution, which leads to the distortion of risk area attribution. i The analysis of the values can identify whether the current identification mechanism is adapted to the real distribution characteristics of geostress in terms of spatial structure, and provide a quantitative basis for dynamically optimizing the risk identification boundary.

[0092] In this embodiment, the logic for obtaining the stress dispersion coefficient of each sub-block is as follows:

[0093] The stress spatial discrete information is extracted from the preprocessed stress structural characteristic information of each sub-block, including the stress value of each monitoring point in each sub-block in the generated regional stress distribution map and the distance from each monitoring point to the geometric center of the sub-block to which it belongs, and calibrated as and represents the stress value of the jth monitoring point in the i-th sub-block in the generated regional stress distribution map, represents the distance from the j-th monitoring point in the i-th sub-block of the generated regional stress distribution map to the geometric center of the sub-block to which it belongs, where i = 1, 2, 3, …, k, j = 1, 2, 3, …, h, and both k and h are positive integers;

[0094] In the generated regional stress distribution map, the system can automatically extract the stress values of the monitoring points in each sub-block and their spatial distances to the geometric centers of the sub-blocks to which they belong through map parsing and spatial data processing modules. The specific acquisition method is as follows: The system first calls the spatial boundary information of the sub-blocks, identifies the set of monitoring points included in each sub-block, and performs spatial numbering and attribution mapping for each monitoring point to form a one-to-one index relationship of "monitoring point - sub-block". Subsequently, the system reads the stress values of each monitoring point in the map. This data is collected from previous sensors and has completed coordinate binding and visualization mapping in the map, so it has clear spatial positioning and quantitative characteristics. These stress values constitute the stress value of the j-th monitoring point in the i-th sub-block Next, the system automatically calculates the coordinates of its centroid or geometric center point according to the geometric boundaries of the sub-blocks, usually the geometric center of the sub-block bounding box or the centroid of the polygon. The system then calculates the actual physical distances from each monitoring point to the center of the sub-block to which it belongs based on the Euclidean distance calculation logic between the three-dimensional coordinates of each monitoring point and the coordinates of the geometric center of the sub-block. These distance values constitute i.e., the distance from the j-th monitoring point to the geometric center of the sub-block. These data extraction and calculation processes can be automatically completed through spatial index processing, coordinate parsing functions, and distance calculation algorithms in the software, with high efficiency and repeatability, and are the basic quantitative support for constructing the stress discrete model of the sub-blocks.

[0095] Calculate the stress dispersion coefficients of each sub-block. The specific calculation formula is as follows:

[0096]

[0097] In the formula, SDC i is the stress dispersion coefficient of the i-th sub-block.

[0098] The calculation formula of this stress dispersion coefficient is designed to quantitatively measure the fluctuation degree and local aggregation characteristics of the in-situ stress distribution within each sub-block, and is the result of weighted modeling of the stress difference and spatial position relationship of the monitoring points. First, represents the deviation between the stress value of the j-th monitoring point in the i-th sub-block and the average stress value of all monitoring points in the sub-block, which is used to measure the deviation degree of this monitoring point in the stress field; the square of this deviation value is used to amplify the influence of larger deviations and at the same time compress the interference of small fluctuations, which is in line with the general practice of variance analysis; secondly, the denominator part It means that the distance from each monitoring point to the geometric center of the sub-block is exponentially processed, and the spatial distance is given a nonlinear weight. The closer the monitoring point is to the center, the greater the impact on the overall discreteness, and the farther the point is, the more the impact is naturally attenuated, thereby achieving enhanced perception of stress disturbances in the central area; then, the weighted square deviation values of all monitoring points are summed, divided by the total number of monitoring points h and squared to obtain the normalized overall spatial discreteness, that is, the stress discreteness coefficient of the sub-block. Through this calculation structure, the system can not only fully reflect the overall volatility of stress values, but also introduce spatial displacement factors, improve the ability to identify local strong disturbances, high-energy aggregation or stress release abnormal areas, and provide a highly reliable physical basis and mathematical support for subsequent identification mechanism evaluation and risk level judgment.

[0099] Stress dispersion coefficient SDC of the i-th sub-block i It reflects whether the stress value distribution between the monitoring points within the sub-block is uniform and whether there is abnormal disturbance. The larger the value, the more violent the stress fluctuation within the sub-block, and the more obvious local high stress concentration or sudden unloading phenomenon. This spatial discreteness is usually strongly correlated with high-risk hidden dangers in the evolution of ground stress. Therefore, when the stress dispersion coefficients of multiple sub-blocks are significantly higher or present a discontinuous distribution state in space, it usually indicates that the currently defined risk identification area fails to effectively cover these high-risk sections with violent fluctuations, indicating that the identification mechanism is not adaptable enough or the identification boundary division is unreasonable. On the contrary, if the stress dispersion coefficients of most sub-blocks are small and the distribution is stable, it means that the stress field is overall balanced, and the current division mechanism can better fit the actual stress distribution characteristics. Therefore, through the statistical and comprehensive evaluation of the stress dispersion coefficients of each sub-block, the adaptability of the current risk identification area division mechanism to the spatial changes of ground stress can be effectively judged, thereby guiding whether the identification strategy needs to be dynamically adjusted.

[0100] In this embodiment, the directional stress gradient index DSGI of each sub-block is generated. i and stress dispersion coefficient SDC i A weighted model is constructed to generate the partition adaptation evaluation coefficient of each sub-block through weighted summation. The specific calculation formula is as follows:

[0101] PAEC i =ω1*DSGI i +ω2*SDC i

[0102] Where, PAEC i is the adaptation evaluation coefficient of the i-th sub-block, ω1 and ω2 are the directional stress gradient index DSGI of each sub-block, respectively. i and stress dispersion coefficient SDC i The non-zero weight coefficient of , and ω1+ω2=1;

[0103] A weighted model is a mathematical modeling method that establishes a linear combination relationship among multiple evaluation factors or characteristic indicators. The purpose is to assign different weights according to the importance of each indicator, so as to comprehensively evaluate the overall performance of a certain target quantity. In this solution, the weighted model is used to combine the directional stress gradient index DSGI i of the i-th sub-block with the stress dispersion coefficient SDC i for numerical fusion to generate the partition adaptation evaluation coefficient PAEC i of this sub-block. This process is controlled by two preset weight coefficients ω1 and ω2 in the system to determine the contribution degree of the two types of parameters in the weighted summation. Among them, ω1 corresponds to the directional stress gradient index and is used to reflect the weight of the stress change intensity of the sub-block in the tunneling direction; ω2 corresponds to the stress dispersion coefficient and is used to reflect the importance of the internal stress fluctuation amplitude of the sub-block in the evaluation. Both of these weight coefficients are positive numbers and satisfy the normalization constraint of ω1 + ω2 = 1. They can be dynamically set according to the geological background, identification experience or machine learning training results. When calculating, the system multiplies DSGI i and SDC i by the corresponding weights and then sums them to obtain the evaluation result of each sub-block, which not only retains the independent characteristics of the two types of indicators but also forms a comprehensive quantitative expression of the partition adaptability. This weighted model can be automatically run through the weight configuration interface and parameter binding mechanism in the software, and has adjustability and adaptability.

[0104] Perform standard deviation analysis on the partition adaptation evaluation coefficient PAEC i of each sub-block to generate a comprehensive evaluation coefficient, according to the formula:

[0105]

[0106] In the formula, CEC is the comprehensive evaluation coefficient, i = 1, 2, 3,..., k, and k is a positive integer.

[0107] In this embodiment, a preset comprehensive evaluation coefficient threshold CEC yuzhi is determined, and after determination, it is compared with the generated comprehensive evaluation coefficient CEC. According to the comparison result, it is evaluated whether the partition mechanism of the current risk identification area is suitable for the current in-situ stress distribution state. The specific comparison and analysis are as follows:

[0108] If CEC ≤ CEC yuzhi , the partition mechanism of the current risk identification area is suitable for the current in-situ stress distribution state;

[0109] This indicates that the fluctuation degree of the division adaptation evaluation coefficients of each sub-block is relatively small as a whole, showing that the system has high consistency and interpretability for the stress distribution characteristics of different sub-blocks. That is, most sub-blocks can be reasonably identified under the current division mechanism, and there are no large-scale local stress evolution mismatches or high-stress aggregation omissions. This shows that the risk identification boundary basically coincides with the actual in-situ stress spatial structure, and the system judgment mechanism has strong adaptability and stability to the spatial gradient, directionality, and local perturbations of the in-situ stress. Therefore, in this case, the division mechanism can maintain the status quo and does not need to be adjusted immediately. It can enter the periodic update or mild optimization mode, which helps to improve the overall operation efficiency of the system and the ability to control response costs. At the same time, it also enhances the credibility of the risk identification results and the accuracy of engineering responses.

[0110] If CEC > CEC yuzhi , the current division mechanism of the risk identification area does not adapt to the current in-situ stress distribution state.

[0111] This situation means that there are obvious fluctuations in the division adaptation evaluation coefficients between sub-blocks, and the response accuracy of the system to different sub-blocks is uneven under the current in-situ stress distribution state. There may be problems such as some high-stress aggregation areas being classified into low-risk areas or stress perturbations not being recognized in time. This evaluation result reflects that the existing boundary division mechanism of the risk identification area fails to effectively capture the non-uniform evolution trend of in-situ stress in space. Especially in the stage where the tunneling disturbance continues to increase, it is extremely easy to cause risks such as fuzzy area identification, misclassification, and early warning lag. Therefore, at this time, the division mechanism must be dynamically adjusted, the risk levels of each sub-block must be re-identified, the boundary range must be corrected, and differential risk response measures must be implemented according to the division results to ensure that the real-time perception ability, risk control ability, and safety early warning effect of the system are not weakened, thereby avoiding the occurrence of sudden disaster events such as rock bursts and roof falls caused by delayed identification.

[0112] According to the evaluation results, each sub-block is divided into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area, and the division boundary of the risk identification area is dynamically adjusted according to the classification results;

[0113] In this embodiment, when the evaluation result is that the current division mechanism of the risk identification area does not adapt to the current in-situ stress distribution state, the pre-set division adaptation evaluation coefficient threshold interval [PAEC min , PAEC max is determined, and after determination, it is compared with the generated division adaptation evaluation coefficient PAEC i of each sub-block. According to the comparison results, each sub-block is divided into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area. The specific division is as follows:

[0114] If PAECi <PAEC min , divide the sub-block into the first recognition and matching area;

[0115] This situation indicates that the directional stress gradient of the sub-block is relatively gentle, the internal stress distribution is relatively uniform, the overall stress state is stable and there is no significant spatial disturbance or concentration trend, indicating that its current risk level determination is highly consistent with the system division mechanism and belongs to the area that the system can accurately identify and does not require further adjustment. Such sub-blocks are divided into the first recognition and matching area, which means that the current recognition mechanism has good adaptability and discriminability in this area. Therefore, the system can maintain its original boundary and risk level division state in the subsequent cycle without triggering additional recognition or response actions, thereby optimizing the allocation of recognition resources, reducing the computing cost, and improving the overall operation efficiency and determination credibility of the system.

[0116] If PAEC min ≤PAEC i ≤PAEC max , divide the sub-block into the second recognition deviation area;

[0117] This situation indicates that there is a certain degree of stress imbalance in the sub-block, such as a slight increase in the directional gradient or a slight expansion of the internal fluctuation, but it has not exceeded the tolerance adaptability error limit of the system. Such areas are divided into the second recognition deviation area, which means that there is a slight deviation in the current division mechanism's recognition of this area, and there is a risk of "boundary ambiguity" or "critical misjudgment". If continuous tunneling or external disturbance increases, it may quickly evolve into a high-risk area. Therefore, the system should enter the early warning mode for such blocks, and can improve the system's dynamic perception ability for this block by strengthening the monitoring frequency, temporarily expanding the boundary buffer, adjusting the evaluation sensitivity, etc., and prioritize the review of its boundary division results in the next cycle to ensure that potential risks can be discovered and intervened in advance.

[0118] If PAEC i >PAEC max , divide the sub-block into the third recognition and reconstruction area;

[0119] This situation indicates that the block shows obvious stress anomaly characteristics, manifested as a sudden increase in the directional stress gradient or severe stress fluctuations, often meaning that there are stress mutation zones, high-energy concentration areas, or stress disturbance instability areas in this region. However, the original division mechanism of the system fails to accurately capture this structural change and may misjudge high-risk areas as medium- and low-risk areas. Such sub-blocks are classified as the third identification and reconstruction area, indicating that the current identification mechanism has seriously mismatched in this area. It is necessary to immediately dynamically reconstruct its identification boundary, risk level, and response strategy, trigger a mandatory risk identification update process, and adjust the risk judgment weight, recalculate the local division logic, or directly intervene in engineering interventions when necessary to avoid major safety hazards caused by lagging risk identification, such as rock bursts, rib spalling, or sudden support failures.

[0120] Determine the threshold interval of the pre-set division adaptation evaluation coefficient, which can be achieved through the historical data modeling and statistical learning module built into the system. The specific methods include clustering analysis and interval estimation based on large-sample operation data. The system first calls the sub-block evaluation sample data under multiple typical tunneling scenarios in the historical monitoring period. These data include the known division adaptation evaluation coefficient PAEC i and its corresponding actual identification accuracy label. The system classifies these PAEC i samples through a clustering analysis algorithm, and identifies the typical numerical distribution intervals corresponding to the three evaluation states of "accurate identification", "identification deviation", and "identification failure". Subsequently, the system extracts the interval boundary values of the evaluation coefficients of each category in the statistical model, and sets a certain confidence level (such as 95%) to exclude the interference of extreme values, so as to determine the minimum and maximum values of the threshold intervals of the division adaptation evaluation coefficients corresponding to the first identification matching area, the second identification deviation area, and the third identification reconstruction area. This interval can be automatically generated by the software, supporting rolling update according to time or dynamic fine-tuning according to the latest identification performance, ensuring the adaptability and discriminability of the evaluation criteria at different tunneling stages and different geological environments, so as to provide a quantitative and sustainable optimization ability numerical basis for sub-block classification.

[0121] Dynamically adjust the division boundary of the risk identification area according to the classification results. Specifically: for the sub-blocks classified as the first identification matching area, keep their original risk identification boundaries unchanged, and continue to use the current identification mechanism for monitoring and evaluation in subsequent cycles; for the sub-blocks classified as the second identification deviation area, set a buffer expansion area on their current boundaries, expand their identification boundary range, and increase the sensitivity parameters of the identification mechanism to enhance the dynamic perception ability of slightly misjudged areas; for the sub-blocks classified as the third identification reconstruction area, reconstruct their original identification boundaries, reset the risk judgment parameters, and enable a dynamic identification mechanism for this sub-block to correct its division boundary in real time to ensure that the risks in this area are accurately identified and effectively covered.

[0122] For the sub-blocks divided into the first recognition matching area, the system can confirm that the sub-blocks show good recognition accuracy and stress stability under the current recognition mechanism by determining that their division adaptation evaluation coefficients are lower than the lower limit of the threshold range. In implementation, the system can mark such blocks as "no adjustment needed", freeze their current boundary data in the recognition boundary maintenance module, and only record the periodic monitoring status and historical fluctuation trends of the sub-blocks for subsequent dynamic comparison. In the calculation of the next cycle, the system can give priority to skipping the boundary reconstruction link and directly inherit the boundary definition of the previous cycle to improve the operation efficiency and reduce the resource consumption of the recognition system. The reason for remaining unchanged is that the stress field characteristics of such areas are highly consistent with the risk assessment results, and further intervention may instead introduce risks of overfitting or misjudgment. Therefore, it is most reasonable to adopt a "stable maintenance" strategy in the control logic, which helps to form the judgment confidence basis of the recognition mechanism.

[0123] For the sub-blocks divided into the second recognition deviation area, the system determines that there are slight errors or boundary deviation risks in their recognition results by comparing that their evaluation coefficients are within the threshold range. At the implementation level, the system can first generate an "expansion buffer zone" automatically within a certain distance outside the current boundary of the sub-block based on the current boundary of the sub-block. The width of this zone can be dynamically adjusted according to historical evaluation errors or the change trend of the current cycle. At the same time, the system will increase the sensitivity parameters in the risk recognition algorithm for this area, such as lowering the stress gradient change threshold, reducing the spatial clustering tolerance, increasing the monitoring frequency, etc., so that the recognition mechanism has a higher response ability to slight stress disturbances. The necessity of such processing lies in that although such areas are not yet high-risk areas, they have already shown structural deviation or trend aggregation characteristics. If not intervened in time, they may quickly evolve into actual risk sources. Therefore, setting a buffer and dynamically perceiving are the key links to ensure early warning.

[0124] For the sub-blocks divided into the third identification and reconstruction area, the system determines that its division adaptation evaluation coefficient significantly exceeds the upper threshold, indicating that the original identification mechanism has seriously failed in this area and cannot accurately reflect its true stress distribution state. In terms of implementation, the system marks this sub-block as "high-risk and needs reconstruction". First, it cancels its original risk boundary, reloads the current stress state data on the original map, and reconstructs its boundary contour with a "new identification logic" by resetting the identification parameters (such as the upper limit of stress fluctuation, local gradient sensitivity, boundary generation algorithm, etc.). At the same time, the system activates a dynamic identification mechanism for this area, allowing the boundary to be continuously corrected in multiple consecutive cycles and performing feedback self-correction in combination with real-time monitoring data. The purpose of doing this is that such blocks are often located in high-stress conversion zones or disturbance transition zones, with drastic and unstable changes. It is necessary to continuously update the identification structure in a fast-adapting manner to ensure that high-risk areas are always within the high-response coverage range of the identification mechanism, so as to achieve precise support and early warning control.

[0125] Based on the divided area types, implement corresponding risk response measures and dynamically correct the risk identification scope based on the response results;

[0126] In this embodiment, based on the divided area types, implement corresponding risk response measures and dynamically correct the risk identification scope, specifically as follows:

[0127] For the sub-blocks divided into the first identification and matching area, maintain the current risk response strategy unchanged, adopt the conventional monitoring frequency and the original support strategy, and record the monitoring data in its subsequent cycles as the reference for evaluation.

[0128] For the sub-blocks divided into the first identification and matching area, since their in-situ stress structures are stable and the evaluation values are lower than the minimum threshold, the system determines them as areas highly adaptable to the identification mechanism. In terms of implementation, the system calls the response strategy parameters of this sub-block in the current cycle, maintains the original support level and monitoring frequency unchanged, and does not perform any active boundary adjustment. At the same time, the system continuously collects the real-time monitoring data (such as stress, microseismic, surrounding rock deformation, etc.) of this area in the background and automatically marks it as "identification reference data" for trend comparison or model fine-tuning verification with other blocks in subsequent cycles. By setting the "no need for active correction" flag in the software platform, it ensures that system resources are concentrated on areas that require more response optimization. The purpose of doing this is to maximize the use of existing identification results, avoid the risk of misidentification caused by repeated evaluation of stable areas, and provide a long-term stable sample for the entire system, improving the model convergence and overall evaluation credibility.

[0129] For the sub-blocks divided into the second identification offset area, an enhanced response strategy is implemented, including increasing the monitoring frequency, introducing auxiliary sensors and early warning algorithms, obtaining dynamic feedback data, and adjusting the boundary shape and coverage ratio of the risk identification scope when there is a deviation between the identification result and the actual response;

[0130] For the sub-blocks divided into the second identification offset area, the system identifies that they are in the "critical adaptation area" of the risk identification mechanism, and there may be mild misjudgment or potential evolution risks. In implementation, the system will increase the monitoring frequency of this block through the parameter regulation module to increase the data sampling density; at the same time, activate the auxiliary early warning algorithms (such as short-term fluctuation capture, micro-perturbation signal analysis) to enhance the real-time perception ability of slight anomalies in this area. If the system finds a deviation between the identification result and the actual risk response (such as abnormal stress increase, local instability event) in two or more consecutive cycles, it will start an automatic correction process, calculate the stress gradient trend of several monitoring points inside and outside the boundary, and accordingly fine-tune the boundary shape or expand the boundary radius to correct the coverage range. This adjustment process is automatically completed by the software in the background without manual intervention. The reason for this is that there is uncertainty in such sub-blocks that "the risk has not been fully released", and by increasing the perception granularity and local boundary adaptability, potential risk evolution can be effectively intervened in advance to avoid response lag when subsequent risk mutations occur.

[0131] For the sub-blocks divided into the third identification reconstruction area, key early warning and active intervention measures are implemented, including real-time monitoring data access, improvement of the support strength level, and resetting of the identification model parameters, and continuously comparing the risk response effect and stress evolution trend in subsequent cycles, and dynamically correcting the risk identification boundary of this sub-block periodically according to the offset result to ensure that this area is always in a high-response control state.

[0132] For the sub-blocks divided into the third identification and reconstruction area, the system recognizes that the original division mechanism can no longer accurately capture the real stress evolution, with serious identification deviations, and the high-response mode needs to be activated immediately. In terms of software implementation, the system will first set this sub-block to the "high-priority processing status", trigger the access of the full-volume real-time monitoring data stream, and temporarily increase the data refresh frequency of the monitoring points. Subsequently, the system activates the high-level support model and the boundary reconstruction engine, redefines the boundary form, updates the risk level label based on the current stress distribution map and its time evolution trend, and at the same time updates the weight parameters and identification thresholds in the identification model. In the subsequent multiple cycles, the system will dynamically compare the actual risk response data (such as surrounding rock failure, energy release events) inside and outside the new boundary range. Once it is found that there are still identification deviations after correction, the identification mechanism parameters will be automatically iterated until the boundary identification is stably convergent. The necessity of this approach is that the stress field evolution in high-risk areas is complex, and static boundary mechanisms often cannot be accurately adapted. It is necessary to form a closed-loop control logic through dynamic model reconstruction and real-time data feedback to ensure "accurate coverage and timely response" of risk control.

[0133] Store the monitoring data, identification results, and response behaviors during the division adjustment process, and continuously optimize the division mechanism of the risk identification area based on the stored historical data to improve the dynamic adaptability and long-term identification accuracy of in-situ stress monitoring.

[0134] To achieve the continuous optimization of the risk identification area division mechanism, the system needs to structurally collect and store the key data during the division adjustment process, mainly including three core data sets: monitoring data (such as stress measurement values, spatial positions, anomaly marks of each sub-block in each cycle), identification results (such as division type, boundary form, evaluation coefficient), and response behaviors (such as adjustment of monitoring frequency, boundary correction amplitude, early warning trigger records, etc.). At the software implementation level, the system packs the above data in real time through the background data link and writes it into the historical record database. Each record contains a timestamp, sub-block identifier, and associated parameters, forming a set of traceable identification and response evolution trajectories. Subsequently, the system calls the built-in analysis engine to perform clustering analysis, trend fitting, and identification model training on the stored data periodically, extracts the effectiveness of various boundary adjustment strategies under different tunneling stages and different identification deviations, and feeds the analysis results back to the model parameter library to dynamically update the identification weights, judgment thresholds, and boundary adaptation criteria, forming a closed-loop of the "data-driven + self-learning" identification mechanism.

[0135] The reason for storing and utilizing the data during the partitioning process is that the in-situ stress field in deep mine shafts has the characteristics of complex spatial structure, drastic temporal variation, and unpredictable local disturbances. Traditional static identification mechanisms are prone to problems such as misjudgment, overfitting, or decreased adaptability during multi-cycle operation. By constructing a complete data chain for the entire process of monitoring - identification - response - correction and continuously injecting it into the optimization process of the identification mechanism, the system can not only gradually build a regional identification experience knowledge base under different geological conditions but also discover the adaptation bottlenecks of the identification mechanism at specific stages and conduct targeted optimization by comparing historical identification results with actual evolution results. At the same time, the data-driven learning process has the ability of non-linear modeling, which can pre-perceive "deviation trends" or "response lag" patterns in future identifications, thereby providing feedforward correction capabilities for the identification logic and significantly enhancing the dynamic adaptability and risk control accuracy of the identification mechanism during long-term operation. This not only ensures the stability and intelligent evolution ability of the system but also provides a continuous and self-updating technical foundation for the entire deep mine shaft monitoring system.

[0136] Such as Figure 2 shown, a real-time in-situ stress monitoring system for deep mine shafts includes a spatial partitioning and modeling module, a stress map generation module, an identification and adaptation evaluation module, a block type discrimination module, a sub-region response correction module, and an identification mechanism optimization module;

[0137] The spatial partitioning and modeling module constructs a spatial sub-block model of the driving area, dynamically partitions the driving area based on the advancing position of the driving face and the distribution information of monitoring points, and establishes a spatial identifier for each partitioned sub-block;

[0138] The stress map generation module collects the in-situ stress monitoring point data within each sub-block and generates a regional stress distribution map based on the spatial relationship between the monitoring points;

[0139] The identification and adaptation evaluation module evaluates whether the partitioning mechanism of the current risk identification area is suitable for the current in-situ stress distribution state based on the generated regional stress distribution map;

[0140] The block type discrimination module divides each sub-block into three categories: the first identification matching area, the second identification deviation area, and the third identification reconstruction area according to the evaluation results, and dynamically adjusts the partitioning boundary of the risk identification area according to the classification results;

[0141] The sub-region response correction module executes corresponding risk response measures based on the divided region types and dynamically corrects the risk identification scope based on the response results;

[0142] The identification mechanism optimization module stores the monitoring data, identification results, and response behaviors during the partitioning adjustment process and continuously optimizes the partitioning mechanism of the risk identification area based on the stored historical data.

[0143] The above formulas are all dimensionless and only take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0144] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired or wireless (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0145] It should be understood that in various embodiments of the present application, the order numbers of the above processes do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0146] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0147] In several embodiments provided by the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the above-described embodiments are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in electrical, mechanical or other forms.

[0148] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0149] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0150] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A real-time monitoring method for in-situ stress in deep mine shafts, characterized in that, Specifically, it includes the following steps: Construct a spatial sub-block model of the tunneling area, dynamically divide the tunneling area based on the advancing position of the tunneling face and the distribution information of monitoring points, and establish a spatial identifier for each divided sub-block; Collect the data of in-situ stress monitoring points in each sub-block, and generate a regional stress distribution map based on the spatial relationship between the monitoring points; Based on the generated regional stress distribution map, evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state; According to the evaluation results, divide each sub-block into three categories: the first identification matching area, the second identification offset area, and the third identification reconstruction area, and dynamically adjust the division boundary of the risk identification area according to the classification results; Based on the divided area types, implement corresponding risk response measures, and dynamically correct the risk identification scope based on the response results; Store the monitoring data, identification results, and response behaviors during the division adjustment process, and continuously optimize the division mechanism of the risk identification area based on the stored historical data.

2. The real-time monitoring method for in-situ stress in deep mine shafts according to claim 1, characterized in that, Collect the data of in-situ stress monitoring points in each sub-block, and generate a regional stress distribution map based on the spatial relationship between the monitoring points. Specifically: collect the real-time stress monitoring values and their three-dimensional spatial coordinate information of all in-situ stress monitoring points located in each sub-block, calculate the stress difference and spatial distance between adjacent monitoring points, construct a polygon topological structure according to the spatial relationship of each monitoring point, and perform interpolation fitting processing on the monitoring point data within this topological structure, and further perform a grid operation to generate a continuous regional stress distribution map, which is used to characterize the trend of in-situ stress gradient change between different sub-blocks in the tunneling direction.

3. The real-time in-situ stress monitoring method for deep mines according to claim 2, characterized in that Based on the generated regional stress distribution map, evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state. Specifically, it includes the following steps: Based on the generated regional stress distribution map, extract the stress structure feature information of each sub-block and perform preprocessing on it; Extract the directional stress distribution information and stress spatial discretization information from the preprocessed stress structure feature information of each sub-block, analyze them, and generate the directional stress gradient index and stress discretization coefficient of each sub-block respectively; Construct a weighted model for the generated directional stress gradient index and stress discretization coefficient of each sub-block, generate a division adaptation evaluation coefficient for each sub-block through weighted summation, and generate a comprehensive evaluation coefficient through the standard deviation calculation formula; Determine a preset threshold for the comprehensive evaluation coefficient, and compare it with the generated comprehensive evaluation coefficient after determination, and evaluate whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state according to the comparison result.

4. A real-time monitoring method for in-situ stress in deep mines according to claim 3, characterized in that, The acquisition logic of the directional stress gradient index of each sub-block is as follows: Extract the directional stress distribution information from the stress structure characteristic information of each preprocessed sub-block, specifically including the average stress value of the front monitoring point group, the average stress value of the rear monitoring point group, and the driving direction length of each sub-block in the generated regional stress distribution map, and calibrate them as QJP i 、HJP i and L i ,QJP i represents the average stress value of the front monitoring point group of the i-th sub-block in the generated regional stress distribution map, HJP i represents the average stress value of the rear monitoring point group of the i-th sub-block in the generated regional stress distribution map, L i represents the driving direction length of the i-th sub-block in the generated regional stress distribution map, i = 1, 2, 3, …, k, where k is a positive integer; Calculate the directional stress gradient index of each sub-block. The specific calculation formula is as follows: In the formula, DSGI i is the directional stress gradient index of the i-th sub-block, and QJP i represents the average value of the stress values of all the monitoring points at the front boundary of the driving direction of the i-th sub-block in the generated regional stress distribution map, and HJP i represents the average value of the stress values of all the monitoring points at the rear boundary of the i-th sub-block in the generated regional stress distribution map, and L i represents the physical length of the i-th sub-block in the driving direction in the generated regional stress distribution map.

5. A real-time monitoring method for in-situ stress in deep mines according to claim 4, characterized in that The acquisition logic of the stress discretization coefficient of each sub-block is as follows: Extract the stress space discretization information from the stress structure characteristic information of each preprocessed sub-block, specifically including the stress values of each monitoring point in each sub-block in the generated regional stress distribution map and the distances from each monitoring point to the geometric center of the sub-block to which it belongs, and calibrate them respectively as and represents the stress value of the j-th monitoring point in the i-th sub-block in the generated regional stress distribution map, represents the distance from the j-th monitoring point in the i-th sub-block in the generated regional stress distribution map to the geometric center of the sub-block to which it belongs, i = 1, 2, 3, …, k, j = 1, 2, 3, …, h, and both k and h are positive integers; Calculate the stress discretization coefficient of each sub-block. The specific calculation formula is as follows: where SDC i is the stress dispersion coefficient of the i-th sub-block.

6. The real-time in-situ stress monitoring method for deep mines according to claim 5, characterized in that, The direction stress gradient index DSGI of each generated sub-block i and the stress dispersion coefficient SDC i Construct a weighted model, and generate a division adaptation evaluation coefficient for each sub-block through weighted summation. The specific calculation formula is as follows: PAEC i = ω1 * DSGI i + ω2 * SDC i wherein, PAEC i is the division adaptation evaluation coefficient of the i-th sub-block, and ω1 and ω2 are the non-zero weight coefficients of the direction stress gradient index DSGI i and the stress dispersion coefficient SDC i respectively, and ω1 + ω2 = 1; The division of each sub-block adapts to the evaluation coefficient PAEC i Perform standard deviation analysis to generate a comprehensive evaluation coefficient according to the formula: In the formula, CEC is the comprehensive evaluation coefficient, i = 1, 2, 3,..., k, and k is a positive integer.

7. A real-time monitoring method for in-situ stress in deep mines according to claim 6, characterized in that, Determine the preset comprehensive evaluation coefficient threshold CEC yuzhi , and compare it with the generated comprehensive evaluation coefficient CEC after determination. According to the comparison result, evaluate whether the current risk identification area division mechanism is suitable for the current in-situ stress distribution state. The specific comparison and analysis are as follows: If CEC ≤ CEC yuzhi , the division mechanism of the current risk identification area adapts to the current in-situ stress distribution state; If CEC > CEC yuzhi , the current division mechanism of the risk identification area does not adapt to the current in-situ stress distribution state.

8. A real-time monitoring method for in-situ stress in deep mines according to claim 7, characterized in that, If the evaluation result shows that the division mechanism of the current risk identification area does not adapt to the current in-situ stress distribution state, determine the pre-set threshold interval of the division adaptation evaluation coefficient [PAEC min , PAEC max , and after determination, compare it with the division adaptation evaluation coefficients PAEC i of each generated sub-block. According to the comparison results, divide each sub-block into three categories: the first recognition matching area, the second recognition offset area, and the third recognition reconstruction area. The specific division is as follows: If PAEC i <PAEC min , divide this sub-block into a first recognition and matching area; If PAEC min ≤PAEC i ≤PAEC max , divide this sub-block into a second recognition offset area; If PAEC i >PAEC max , divide this sub-block into a third recognition and reconstruction area; Dynamically adjust the division boundary of the risk identification area according to the classification results, specifically as follows: For the sub-blocks classified as the first identification matching area, keep their original risk identification boundaries unchanged, and continue to use the current identification mechanism for monitoring and evaluation in subsequent cycles; For the sub-blocks classified as the second identification deviation area, set a buffer expansion area on their current boundaries, expand their identification boundary ranges, and increase the sensitivity parameters of the identification mechanism; For the sub-blocks classified as the third identification reconstruction area, reconstruct their original identification boundaries, reset the risk judgment parameters, and enable a dynamic identification mechanism for this sub-block to real-time correct its division boundary.

9. A real-time monitoring method for in-situ stress in deep mines according to claim 8, characterized in that, Based on the divided area types, execute corresponding risk response measures, and dynamically correct the risk identification scope based on the response results, specifically as follows: For the sub-blocks classified as the first identification matching area, maintain the current risk response strategy unchanged, adopt the conventional monitoring frequency and the original support strategy, and record the monitoring data in subsequent cycles as the reference for evaluation; For the sub-blocks classified as the second identification deviation area, execute an enhanced response strategy, including increasing the monitoring frequency, introducing auxiliary sensors and warning algorithms, obtaining dynamic feedback data, and adjusting the boundary shape and coverage ratio of the risk identification scope when there is a deviation between the identification result and the actual response; For the sub-blocks classified as the third identification reconstruction area, execute key warning and active intervention measures, including real-time monitoring data access, improvement of the support strength level, and reset of the identification model parameters, and continuously compare the risk response effect and the stress evolution trend in subsequent cycles, and dynamically correct the risk identification boundary of this sub-block periodically according to the deviation result.

10. A real-time in-situ stress monitoring system for deep mines, which is used to implement the real-time in-situ stress monitoring method for deep mines described in any one of the above claims 1-9, characterized in that, Including a spatial division modeling module, a stress map generation module, an identification adaptation evaluation module, a block type discrimination module, a partition response correction module, and an identification mechanism optimization module; The spatial division modeling module constructs a spatial sub-block model of the tunneling area, dynamically divides the tunneling area based on the advancing position of the tunneling face and the distribution information of the monitoring points, and establishes a spatial identifier for each divided sub-block; The stress map generation module collects the data of the in-situ stress monitoring points in each sub-block, and generates a regional stress distribution map based on the spatial relationship between the monitoring points; The identification adaptation evaluation module evaluates whether the division mechanism of the current risk identification area is suitable for the current in-situ stress distribution state based on the generated regional stress distribution map; The block type discrimination module divides each sub-block into three categories: the first identification matching area, the second identification deviation area, and the third identification reconstruction area according to the evaluation results, and dynamically adjusts the division boundary of the risk identification area according to the classification results; The partition response correction module executes corresponding risk response measures based on the divided area types, and dynamically corrects the risk identification scope based on the response results; The identification mechanism optimization module stores the monitoring data, identification results, and response behaviors during the division adjustment process, and continuously optimizes the division mechanism of the risk identification area based on the stored historical data.

Images