Anchor Bolt Stress Analysis Method Based on Monitoring Time Series Data

By constructing an anchor bolt stress response field and dynamic early warning threshold, and combining stress-displacement correlation analysis, the problems of insufficient spatial resolution and rigid early warning threshold in existing anchor bolt stress analysis methods are solved. This enables full-domain stress situation perception and categorized hierarchical early warning, thereby improving the refined control capability of surrounding rock safety monitoring.

CN123116003APending Publication Date: 2026-11-17JILIN WATER NETWORK DEVELOPMENT GROUP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611263506.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing anchor stress analysis methods cannot reflect the full spatial distribution of anchor support stress in the spatial dimension. The early warning threshold has poor adaptability, cannot be dynamically adjusted, and lacks multi-source data correlation analysis, resulting in inaccurate early warning results and a high risk of false alarms. This makes it difficult to meet the needs of refined safety management of surrounding rock under deep and complex geological conditions.

Method used

Based on monitoring time-series data, by collecting data from anchor bolt stress gauges and multi-point displacement gauges, and after removing invalid data, an anchor bolt stress response field is constructed, the correlation between stress and displacement is calculated, early warning thresholds are dynamically generated, and graded early warnings are carried out in combination with stress response zoning information. Kriging interpolation is used for spatial interpolation and early warning determination.

Benefits of technology

It realizes the mapping from single-point monitoring to the whole-domain stress response field, accurately identifies high-stress response areas and low-stress response areas, enhances the engineering guidance of early warning results, reduces false alarm rate, and improves the response speed and accuracy of surrounding rock safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116003A_ABST
    Figure CN123116003A_ABST
Patent Text Reader

Abstract

This invention discloses a method for anchor bolt stress analysis based on monitoring time-series data, comprising the following steps: collecting time-series monitoring data from anchor bolt stress gauges and multi-point displacement gauges on the same cross-section, and removing invalid data; extracting stress statistical features according to a first preset time window, constructing an anchor bolt stress response field by interpolating the spatial distribution of measuring points, identifying high-stress response areas and low-stress response areas, and outputting zoning information; aligning stress and displacement time-series data, calculating the correlation between the two to obtain correlation feature parameters; dynamically generating early warning thresholds based on the stress change trend of a second time window, and comprehensively judging by combining zoning information and correlation parameters to generate graded early warning signals; and outputting analysis results including measuring point locations and early warning levels. This invention solves the problems of insufficient spatial resolution of single-point monitoring, poor adaptability of early warning thresholds, difficulty in distinguishing the causes of anomalies, and insufficient accuracy and engineering guidance in traditional anchor bolt stress analysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the technical field of safety monitoring of surrounding rock in underground engineering, and specifically to a method for analyzing anchor stress based on monitoring time-series data. Background Technology

[0002] With the increasing scale and depth of underground engineering projects such as tunnels, underground cavern complexes, and mine roadways, the evaluation of surrounding rock stability and the prevention and control of geological disasters have become core aspects of engineering safety management. As a core component of active rock support, the stress state of anchor bolts is a key indicator reflecting the stress evolution of the surrounding rock and evaluating the effectiveness of the support system. Therefore, anchor bolt stress monitoring has become an essential part of the safety monitoring system for various underground engineering projects.

[0003] Currently, conventional anchor bolt stress analysis is mostly based on independent monitoring data from discrete measuring points, directly comparing real-time stress measurements with preset fixed thresholds to achieve over-limit warnings. This approach has several shortcomings in practical applications: Spatially, limited by the number and location of measuring points, the monitoring data can only reflect the local stress state at a single point, failing to depict the overall spatial distribution pattern of anchor bolt support stress. It is difficult to accurately identify contiguous high-stress, high-risk areas and low-stress, loosening areas, easily leading to blind spots where some measuring points are normal while regional risks have already accumulated. Overall spatial resolution and situational awareness are insufficient. Regarding warning mechanisms, existing schemes generally use fixed warning thresholds, unable to dynamically adjust the judgment criteria according to the recent evolution trend of anchor bolt stress. During periods of stable stress, frequent false alarms are easily triggered by normal fluctuations, while during periods of rapid stress increase, there are problems of delayed response and untimely warnings, making it difficult to adapt to the actual dynamic changes in anchor bolt stress. In terms of analytical dimensions, most methods rely solely on a single stress indicator for judgment, failing to effectively combine cross-sectional displacement monitoring data for multi-source correlation analysis. This makes it impossible to classify anomalies or assess their tendencies. Early warning results only provide over-limit alerts, lacking classification and qualitative capabilities and engineering guidance significance, making it difficult to directly support on-site tiered response decisions. At the data processing level, there is a lack of a systematic pre-processing mechanism for invalid data from sensor malfunctions, environmental interference, and construction disturbances. This can easily lead to biased subsequent analysis results and false warnings due to contaminated data, affecting the overall reliability of the analysis results.

[0004] In summary, existing anchor bolt stress analysis methods suffer from insufficient spatial perception capabilities, poor adaptability of early warning thresholds, low differentiation of anomaly types, and weak data quality control, making it difficult to meet the needs of refined safety management of surrounding rock in underground engineering under deep and complex geological conditions. Summary of the Invention

[0005] This invention provides a method for anchor bolt stress analysis based on monitoring time-series data to solve the aforementioned technical problems, specifically employing the following technical solution: The anchor bolt stress analysis method based on monitoring time series data includes the following steps: S1. Collect stress time-series monitoring data from anchor stress gauges and displacement time-series monitoring data from multi-point displacement gauges on the same cross section, and remove invalid data from both types of data; S2. Extract the stress statistical characteristics of the stress time series monitoring data according to the first preset time window, interpolate according to the spatial distribution of the anchor stress gauge, construct the anchor stress response field, identify the high stress response area and low stress response area according to the anchor stress response field, and output the anchor stress response zoning information. S3. Align the stress time-series monitoring data and the displacement time-series monitoring data according to the timestamp, calculate the correlation between the stress measurement value and the displacement measurement value at the same moment, and obtain the correlation characteristic parameters; S4. Based on the changing trend of the stress time-series monitoring data within the second preset time window, a warning judgment threshold is dynamically generated. If the real-time stress measurement value exceeds the warning judgment threshold, a comprehensive judgment is made by combining the anchor stress response zoning information and the associated feature parameters to generate a graded warning signal. S5. Based on the graded early warning signal, output the anchor bolt stress analysis results, including the location of the early warning measuring point and the early warning level.

[0006] Further, the step of extracting the stress statistical features of the stress time-series monitoring data according to the first preset time window includes: Set the window length and sliding step size of the first preset time window, and sequentially truncate the stress time series monitoring data according to the sliding step size based on the window length to obtain multiple stress time series segments; The stress timing segments are sequentially continuous and adjacent segments partially overlap. The arithmetic mean, maximum, minimum, median, and standard deviation of the stress measurements within the stress time series segment are calculated and used as the stress statistical feature vector of the stress time series segment.

[0007] Furthermore, the step of interpolating based on the spatial distribution of the anchor bolt stress gauges to construct the anchor bolt stress response field includes: Using the three-dimensional coordinates of the anchor stress gauge in the spatial coordinate system of the cavern group as interpolation sample points, and the stress statistical feature vector corresponding to the first preset time window of the interpolation sample points as interpolation attributes, an interpolation grid is laid out within the spatial range of the surrounding rock of the cavern group. The stress statistical eigenvectors at the grid nodes are estimated unbiased optimally using the Kriging interpolation method. The average value of the stress statistical feature vector at the grid node is extracted as the stress amplitude. Combined with the grid node coordinates and the first preset time window index, the anchor stress response field is generated.

[0008] Furthermore, the method of using Kriging interpolation to perform unbiased optimal estimation of the stress statistical eigenvectors at the grid nodes includes: The spatial distance between any two points is calculated based on the three-dimensional coordinates of the interpolated sample points. The attribute difference is calculated for each statistical component of the feature vector. A semi-variogram model is fitted based on the correspondence between the spatial distance and the attribute difference. The Kriging weight coefficients between the grid nodes and the interpolated sample points are determined based on the semi-variogram model. Based on the Kriging weight coefficients, the stress statistical feature vectors of the interpolated sample points are weighted and summed to obtain the stress statistical feature vectors of the grid nodes.

[0009] Further, the step of identifying high-stress response regions and low-stress response regions based on the anchor bolt stress response field and outputting anchor bolt stress response zoning information includes: Locate the local maximum and local minimum points of stress amplitude at each spatial location in the anchor bolt stress response field; Using the local maximum point as the center and the stress amplitude isosurface at that point as the boundary, the high stress response influence range is defined, and the region within the high stress response influence range where the stress amplitude exceeds a preset first threshold is marked as the high stress response region. Using the local minimum point as the center and the stress amplitude isosurface at that point as the boundary, the low stress response influence range is defined, and the region within the low stress response influence range where the stress amplitude is lower than a preset second threshold is marked as the low stress response region. All marked high-stress response areas and low-stress response areas are integrated into anchor bolt stress response zoning information.

[0010] Furthermore, the calculation of the correlation between stress measurements and displacement measurements at the same time, yielding correlation characteristic parameters, includes: At the same time, stress measurements and displacement measurements that match the spatial location are paired as monitoring data pairs; Normalization is performed on the stress and displacement measurements in the monitoring data pairs respectively; Calculate the Pearson correlation coefficient of the normalized monitoring data pairs, and use the Pearson correlation coefficient as the correlation feature parameter.

[0011] Furthermore, the step of dynamically generating a warning threshold based on the changing trend of the stress time-series monitoring data within a second preset time window includes: Obtain the stress measurement values ​​at each acquisition time within the second preset time window; The stress change rate is calculated based on the stress measurements at the start and end times of the second preset time window, and the stress fluctuation amplitude is calculated based on the stress measurements at each acquisition time within the second preset time window. If the stress change rate does not exceed the preset change rate threshold, the initial stress value is linearly adjusted based on the stress fluctuation amplitude to generate an early warning judgment threshold. If the stress change rate exceeds the preset change rate threshold, the initial stress value is nonlinearly adjusted based on the stress change rate and the stress fluctuation amplitude to generate a warning judgment threshold.

[0012] Furthermore, if the real-time stress measurement value exceeds the warning judgment threshold, a comprehensive judgment is made by combining the anchor bolt stress response zoning information and the associated feature parameters to generate a graded warning signal, including: When the real-time stress measurement value exceeds the warning judgment threshold, the spatial coordinates of the anchor stress gauge corresponding to the real-time stress measurement value are marked as the abnormal measurement point location. Obtain the anchor stress response zoning at the abnormal measurement point location and the corresponding associated characteristic parameters; If the location of the abnormal measuring point is within the high-stress response area and the associated characteristic parameters indicate a cooperative change relationship, then the overall deformation of the surrounding rock is determined to be abnormal, and a first-level early warning signal is generated. If the location of the abnormal measuring point is within the high-stress response area and the associated characteristic parameters indicate that there is no coordinated change relationship, then a local geological impact anomaly is determined, and a secondary early warning signal is generated. If the location of the abnormal measuring point is not located within the high stress response area and the associated characteristic parameters indicate a cooperative change relationship, then the anomaly is determined to be a deformation-affected anomaly in the adjacent area, and a secondary early warning signal is generated. If the location of the abnormal measuring point is not within the high-stress response area and the associated characteristic parameters indicate no coordinated change relationship, then a local disturbance anomaly is determined, and a level three warning signal is generated.

[0013] Furthermore, the step of outputting anchor bolt stress analysis results, including the location of the warning measurement point and the warning level, based on the graded warning signal, includes: The graded early warning signal is analyzed to extract the early warning level identifier and the device number of the anchor stress gauge that issued the graded early warning signal. Based on the equipment number, retrieve the pre-stored equipment ledger and retrieve the installation coordinates of the anchor stress gauge corresponding to the equipment number as the location of the early warning measurement point; Convert the warning level identifier into a corresponding warning level text description; Write the location of the early warning measurement point and the text description of the early warning level into the corresponding fields of the analysis result record table to generate the anchor bolt stress analysis results.

[0014] Furthermore, after generating the anchor bolt stress analysis results, the process also includes: The verification range is defined centered on the location of the warning measurement point with a warning level of Level 1 in the anchor bolt stress analysis results; Extract stress time-series monitoring data of all anchor stress gauges within the second preset time window within the review scope, and sequentially perform anchor stress response field construction, anchor stress response partition identification, related feature parameter calculation, and graded early warning signal generation on the stress time-series monitoring data to obtain the review judgment result; The verification result is written into the anchor bolt stress analysis result.

[0015] The advantages of this invention lie in its ability to extend discrete single-point anchor stress monitoring into a continuous anchor stress response field, accurately identifying high-stress response high-risk areas and low-stress response loosening areas, thus overcoming the shortcomings of traditional single-point monitoring in terms of spatial perception. By combining cross-sectional displacement monitoring data to quantify the stress-displacement correlation, it can effectively classify different anomaly types such as overall rock deformation, local geological influence, adjacent area deformation influence, and local disturbance, enhancing the engineering guidance of the early warning results. The invention employs a trend-adaptive dynamic early warning threshold and a multi-factor comprehensive grading and judgment mechanism, balancing early warning sensitivity and accuracy, reducing false alarm rates while improving the response speed to rapid instability risks. A supporting first-level early warning verification mechanism further ensures the reliability of high-risk early warnings, providing strong support for the refined management of surrounding rock safety in underground engineering. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the anchor bolt stress analysis method based on monitoring time series data of the present invention; Figure 2 This is a schematic diagram of the system in which the anchor stress analysis method based on monitoring time series data of the present invention is applied; Detailed Implementation

[0018] Embodiments of the present invention are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0019] like Figure 1 The method for anchor bolt stress analysis based on monitoring time series data according to this application is shown, which includes the following steps: S1. Collect stress time-series monitoring data from anchor stress gauges and displacement time-series monitoring data from multi-point displacement gauges on the same cross section, and remove invalid data from both types of data; S2. Extract the stress statistical characteristics of the stress time series monitoring data according to the first preset time window, interpolate according to the spatial distribution of the anchor stress gauge, construct the anchor stress response field, identify the high stress response area and low stress response area according to the anchor stress response field, and output the anchor stress response zoning information. S3. Align the stress time-series monitoring data and the displacement time-series monitoring data according to the timestamp, calculate the correlation between the stress measurement value and the displacement measurement value at the same moment, and obtain the correlation characteristic parameters; S4. Based on the changing trend of the stress time-series monitoring data within the second preset time window, a warning judgment threshold is dynamically generated. If the real-time stress measurement value exceeds the warning judgment threshold, a comprehensive judgment is made by combining the anchor stress response zoning information and the associated feature parameters to generate a graded warning signal. S5. Based on the graded early warning signal, output the anchor bolt stress analysis results, including the location of the early warning measuring point and the early warning level.

[0020] The anchor stress analysis method based on monitoring time-series data provided by this invention can be applied to anchor stress monitoring and surrounding rock stability analysis in geotechnical engineering scenarios such as tunnel engineering, underground cavern groups, and mine roadways. It is compatible with monitoring data processing from various types of anchor stress gauges, including vibrating wire and fiber optic types, and multi-point position gauges. This method uses anchor stress time-series monitoring data as its core, combined with same-section displacement monitoring data, to construct a complete processing chain including sliding window feature extraction, anchor stress response field construction, stress-displacement correlation, dynamic threshold early warning, and multi-factor hierarchical judgment. This achieves an upgrade from single-point stress monitoring to global anchor stress situation awareness, and from fixed threshold alarms to categorized hierarchical early warning, effectively solving the problems of insufficient spatial resolution, rigid early warning thresholds, and low anomaly type differentiation in traditional anchor stress analysis methods. The steps described above are detailed below.

[0021] For step S1, stress time-series monitoring data from anchor stress gauges and displacement time-series monitoring data from multi-point displacement gauges on the same cross section are collected, and invalid data in both types of data are removed.

[0022] Specifically, in target engineering areas such as tunnels, underground cavern complexes, and mine roadways, anchor stress gauges and multi-point displacement gauges are synchronously deployed along the surrounding rock monitoring sections according to the designed burial depth and spacing. The two types of sensors are arranged one-to-one at the same surrounding rock depth on the same section, ensuring that they monitor the mechanical response of the surrounding rock in the same area. The anchor stress gauges continuously collect time-series monitoring data of the stress corresponding to the anchor bolt's force in the surrounding rock, while the multi-point displacement gauges synchronously collect time-series monitoring data of displacement within the surrounding rock. Both types of sensors are connected to a unified acquisition clock and use a consistent sampling frequency, which can be ≤0.5 times / h depending on the engineering monitoring level. s Flexible configuration within the range of ≤2 times / h; this embodiment preferably uses a sampling frequency f. s =1 time / hour, ensuring the time alignment accuracy of the two types of data from the acquisition end, and providing a unified time reference for subsequent time series correlation analysis.

[0023] After the initial monitoring data acquisition is completed, a three-tiered verification mechanism is used to systematically filter and eliminate invalid data from both types of data. First, a range verification is performed, directly identifying and eliminating any measurements exceeding the sensor's nominal measurement range. For example, for a vibrating wire anchor stress gauge with a nominal range of 0-30 MPa, data exceeding the limit (below 0 MPa or above 30 MPa) is automatically filtered out. This verification directly eliminates completely distorted data caused by sensor over-range damage or data overflow from the acquisition module, safeguarding the first line of defense for data validity at the most basic physical boundary level. Next, a constant value verification is performed, iterating through the measurement sequence of consecutive acquisition cycles. If the measurement value remains completely constant and without any fluctuation for three or more consecutive acquisition cycles, this segment of data is identified as invalid and eliminated. This verification effectively identifies "dead data" caused by faults such as broken sensor signal lines, jammed acquisition channels, or interrupted data transmission, preventing stagnant and false monitoring values ​​from entering subsequent analysis stages and causing erroneous judgments of stress stability. Finally, a sudden change verification is performed, calculating the change in measured values ​​between two adjacent acquisition times point by point. When the stress change exceeds 5 MPa or the displacement change exceeds 10 mm within a single cycle, the jump point is determined as invalid data and discarded. This threshold is a typical engineering experience value and can be adjusted according to the sensor range and engineering geological conditions. This verification can filter out spike jump noise caused by on-site blasting impacts, strong electromagnetic interference, and instantaneous failures of acquisition equipment, eliminating the interference of instantaneous outliers on subsequent statistical feature calculations and trend analysis, and avoiding false stress change warnings caused by single-point jumps.

[0024] The above steps, through a synchronous acquisition scheme and a three-level progressive invalid data removal mechanism, complete the quality control of multi-source monitoring data at the data input end. This ensures the matching of stress and displacement data in terms of spatial location and time scale, and systematically filters out various invalid data caused by equipment failure, environmental interference, and transmission anomalies. This provides a clean and reliable basic data source for subsequent stress statistical feature extraction, anchor bolt stress response field construction, stress-displacement correlation calculation, and graded early warning judgment. It reduces the risk of analytical bias and false alarms caused by invalid data from the source, and ensures the accuracy and engineering credibility of the output results of the entire anchor bolt stress analysis method.

[0025] For step S2, the stress statistical characteristics of the stress time series monitoring data are extracted according to the first preset time window, and interpolation is performed according to the spatial distribution of the anchor stress gauge to construct the anchor stress response field. Based on the anchor stress response field, high stress response areas and low stress response areas are identified, and anchor stress response zoning information is output.

[0026] The step of extracting stress statistical features from the stress time-series monitoring data according to the first preset time window includes: setting the window length and sliding step size of the first preset time window; sequentially truncating the stress time-series monitoring data according to the sliding step size using the window length to obtain multiple stress time-series segments; the stress time-series segments are sequentially continuous and adjacent segments partially overlap; calculating the arithmetic mean, maximum, minimum, median, and standard deviation of the stress measurements within the stress time-series segments as the stress statistical feature vector of the stress time-series segments.

[0027] The step of interpolating based on the spatial distribution of the anchor stress gauges to construct the anchor stress response field includes: using the three-dimensional coordinates of the anchor stress gauges in the spatial coordinate system of the cavern group as interpolation sample points, and using the stress statistical feature vector corresponding to the first preset time window of the interpolation sample points as the interpolation attribute, and laying out an interpolation grid within the spatial range of the surrounding rock of the cavern group; using the Kriging interpolation method to perform unbiased optimal estimation of the stress statistical feature vector at the grid nodes; extracting the average value of the stress statistical feature vector at the grid nodes as the stress amplitude, and combining the grid node coordinates and the index of the first preset time window to generate the anchor stress response field.

[0028] The method of using Kriging interpolation to perform unbiased optimal estimation of the stress statistical feature vector at the grid node includes: calculating the spatial distance between any two points based on the three-dimensional coordinates of the interpolation sample points; calculating the attribute difference for each statistical component of the feature vector; fitting a semi-variogram model based on the correspondence between the spatial distance and the attribute difference; determining the Kriging weight coefficient between the grid node and the interpolation sample point based on the semi-variogram model; and performing a weighted summation of the stress statistical feature vector of the interpolation sample point based on the Kriging weight coefficient to obtain the stress statistical feature vector of the grid node.

[0029] The step of identifying high-stress response regions and low-stress response regions based on the anchor bolt stress response field and outputting anchor bolt stress response zoning information includes: locating local maxima and local minima of stress amplitude at various spatial locations in the anchor bolt stress response field; defining a high-stress response influence range centered on the local maxima and bounded by the stress amplitude isosurface at that point, and marking regions within the high-stress response influence range where the stress amplitude exceeds a preset first threshold as high-stress response regions; defining a low-stress response influence range centered on the local minima and bounded by the stress amplitude isosurface at that point, and marking regions within the low-stress response influence range where the stress amplitude is below a preset second threshold as low-stress response regions; and integrating all marked high-stress response regions and low-stress response regions into anchor bolt stress response zoning information.

[0030] Specifically, the first preset time window is first set to a window length of T1 and a sliding step size of Δt1. The window length can be configured within the range of 12h≤T1≤72h according to the evolution rate of the surrounding rock stress. The sliding step size is usually taken as 1 / 4 to 1 / 2 of the window length. In this embodiment, T1=24h and Δt1=6h are taken. The window length is used to sequentially translate and extract stress time series monitoring data along the preprocessed stress time series data, starting from the first valid data time. Each translation generates an independent stress time series segment. Finally, a set of multiple time series continuous segments with partial overlap of adjacent segments is obtained. The overlapping window design can ensure the time series continuity of adjacent analysis periods, avoid the stress change characteristics at the window boundary being cut off, improve the time resolution of time series analysis, and reduce feature jumps caused by the sliding window.

[0031] For each stress time series segment, multiple statistical indices of the stress measurements within that segment are calculated sequentially and combined to form a stress statistical feature vector, where the feature vector F corresponding to the i-th stress time series segment is... i Recorded as: , in, The arithmetic mean of the measured stress values ​​within the segment is calculated using the following formula: , In the formula, n is the number of valid measurement points within the sub-segment, and σ i,k This is the k-th stress measurement value within the sub-segment.

[0032] σ max,i With σ min,i These represent the maximum and minimum values ​​of the stress measurement within the sub-segment, respectively. Together, they characterize the extreme range of stress fluctuations within that time period, and can be used to identify abnormal periods of sudden stress rises and falls; σ med,i The median value of the stress measurements within the segment is less affected by extreme outliers than the average value, and can serve as a robust reference for stress levels, reducing the interference of residual outlier data on statistical results; σ std,i The standard deviation of the stress measurements within the segment is calculated using the following formula: , Standard deviation is used to quantify the severity of stress fluctuations within a given period; a larger value indicates a more unstable stress state and can serve as an auxiliary indicator for discriminating changes in the mechanical state of surrounding rock. By replacing single instantaneous measurements with feature vectors composed of multi-dimensional statistical features, a comprehensive coverage of four characteristics—overall stress level, extreme range, robust baseline, and degree of fluctuation—is achieved. This not only enhances the richness of feature representation but also mitigates the impact of single-frame monitoring noise through statistical averaging, providing more stable and complete attribute data for subsequent spatial interpolation.

[0033] After extracting the stress statistical features of all measuring points within the corresponding time windows, the first step is to lay out the three-dimensional interpolation grid. Based on the design outline of the cavern group and the influence range of the surrounding rock, the spatial bounding box of the surrounding rock analysis area is determined. This involves taking the minimum and maximum values ​​of the x, y, and z coordinates of all anchor stress gauges and extending them outward by a buffer distance of 5-10m to form a cuboid analysis domain covering all monitoring points and the surrounding rock. Then, along the x, y, and z coordinate axes, the analysis domain is uniformly divided with a preset grid step size to generate a layered array of regular three-dimensional grid nodes. Each grid node corresponds to a unique three-dimensional spatial coordinate. In this embodiment, the grid step size is 1m. This scale ensures that the spatial resolution meets the positioning accuracy requirements of high stress response areas in engineering, while preventing a surge in interpolation calculations due to excessive grid density. This allows the calculation time of a single frame of stress field to be controlled within the real-time processing threshold of the engineering monitoring platform. All sample points used for interpolation were obtained from anchor stress gauges deployed on-site. The three-dimensional installation coordinates of each measuring point were accurately measured using a total station during the installation phase and entered into the spatial database. Before interpolation calculation, the spatial coordinates of each measuring point were bound one-to-one with the stress statistical feature vector of the corresponding first time window, forming an interpolation sample set with spatial attributes. The coordinates of the j-th measuring point are denoted as (x... j ,y j ,z jThe corresponding eigenvector is denoted as F. j j=1,2,...,N, where N is the total number of effective stress gauge points participating in the interpolation. By pre-binding spatial coordinates and attribute data, these points can be directly called in subsequent interpolation calculations, improving computational efficiency.

[0034] To quantitatively describe the spatial autocorrelation of stress properties and support subsequent Kriging interpolation calculations, the experimental semivariogram is first calculated based on all sample point pairs. The statistical components of the eigenvector are calculated separately. Taking the mean stress component as an example, the formula for calculating the experimental semivariogram γ*(h) is: , Where h is the spatial distance between two sample points, calculated using the Euclidean distance between the three-dimensional coordinates of the two points, and N(h) is the number of sample point pairs with a spatial distance of h. For position x i The average stress at that location, For position x i The average stress at +h is calculated using the L2 norm operation of the vector. During the calculation, sample point pairs are grouped and statistically analyzed at fixed intervals to eliminate the interference of random fluctuations at single points. This clearly reveals the variation of stress properties with increasing spatial distance, providing a reliable experimental data foundation for subsequent theoretical model fitting.

[0035] After obtaining the discrete experimental semivariogram points, a nonlinear least squares method is used to fit the spherical theoretical semivariogram model. Through iterative optimization, the sum of squared residuals between the calculated and experimental values ​​is minimized, ultimately determining the core parameters of the model. The expression for the fitted spherical model γ(h) is: , Where C0 is the nugget value, corresponding to the variation at zero distance, reflecting the random fluctuations in properties caused by sensor measurement errors and microscale geological bodies; C1 is the arch height, corresponding to the variation increment caused by spatial correlation; and 'a' is the range, representing the maximum distance at which the stress properties exhibit spatial correlation. When the distance between two points exceeds the range, the stress properties no longer have statistical correlation. The spherical model can accurately match the distribution law of stress in rock and soil masses as a function of spatial distance, which first varies slowly, then increases rapidly, and finally tends to stabilize. The model parameters obtained from the fitting can be directly used for subsequent calculation of kriging weights, ensuring that the interpolation results conform to the physical distribution characteristics of the anchor stress response field.

[0036] For each grid node to be estimated, a Kriging equation system is constructed based on a well-fitted semi-variogram model. Only sample points within the range are selected for calculation, and the interpolation weights of each sample point for that node are solved. Let the node to be estimated be x0. Based on the criteria of unbiased estimation and minimum variance, a Lagrange multiplier μ is introduced, and the matrix form of the constructed Kriging linear equation system is as follows: , In the formula, i and j are the index numbers of the sample points, and their values ​​range from 1, 2, ..., N, γ ij Let γ be the element in the i-th row and j-th column of the matrix, corresponding to the semivariogram value between the i-th and j-th sample points, calculated by substituting the spatial distance between the two points into the spherical model. 0j Let λ be the semivariogram value between node x0 and the j-th sample point to be estimated. j Let be the Kriging weight coefficient corresponding to the j-th sample point. By solving this system of equations using Gaussian elimination, the weight values ​​of all sample points can be obtained, satisfying the unbiased condition that the sum of the weights is 1. This solution process can minimize the variance of the interpolation estimate while ensuring unbiased estimation, truly achieving the optimal linear unbiased estimation of spatial attributes.

[0037] After obtaining the weights of each sample point, the estimated value F of the stress statistical eigenvector at that grid node is obtained by weighted summation. * (x0), the calculation formula is: , This weighting method assigns higher weights to sample points that are closer in space and have stronger correlations, and lower weights to sample points that are farther apart. This perfectly matches the continuous and gradual spatial distribution characteristics of the anchor bolt stress response field. Compared with simple interpolation methods such as inverse distance weighting, it offers higher interpolation accuracy and better smoothness. After traversing all grid nodes to complete the feature vector estimation, the average stress value in the feature vector of each node is extracted as the stress amplitude at that location, forming a three-dimensional stress amplitude field corresponding to a single time window. The three-dimensional stress fields corresponding to all time windows are arranged sequentially according to time order, thus forming an anchor bolt stress response field containing spatial coordinates, time index, and stress amplitude. This field data can be directly used for subsequent dynamic visualization and anchor bolt stress response zoning identification, realizing a complete mapping from discrete measurement points to a continuous spatiotemporal field.

[0038] Based on the constructed anchor stress response field, automatic identification of high-stress response regions and low-stress response regions is performed frame by frame. First, local extremum point localization is performed. For each internal mesh node in the current frame's 3D stress field, all neighboring nodes within its 26-neighborhood are traversed, and the stress amplitude is compared. If the stress amplitude of the current node is greater than the amplitudes of all 26 neighboring nodes, it is marked as a local maximum point; if the stress amplitude of the current node is less than the amplitudes of all neighboring nodes, it is marked as a local minimum point. Nodes at the boundary of the analysis domain do not participate in the extremum search, avoiding false extremum misjudgments caused by boundary effects. The 26-neighborhood omnidirectional search method can cover all adjacent positions in 3D space, ensuring that no local stress anomaly points are missed, guaranteeing the comprehensiveness of stress anomaly identification.

[0039] After locating the extreme points, for each local maximum point, the MarchingCubes 3D isosurface extraction algorithm is used to track the closed isosurface corresponding to a preset stress threshold. The spatial region enclosed by this isosurface is taken as the high stress response influence range. Based on this, the region within the influence range where the stress amplitude exceeds a preset first threshold is officially marked as a high stress response region. The first threshold is dynamically determined using the quantile method, that is, the stress amplitude of all effective grid nodes in the current frame is extracted, sorted from smallest to largest, and the 75th percentile value is taken as the first threshold. This threshold can be adaptively adjusted according to the overall stress level of the surrounding rock, avoiding the problem of poor adaptability of a fixed threshold in different construction stages and different geological regions. This ensures that the identified regions have significant stress increase characteristics, and avoids misjudging normal stress areas as concentrated areas due to excessively low thresholds. For each local minimum point, the same isosurface tracking method is used to delineate the low stress response influence range. The region within the range where the stress amplitude is lower than a preset second threshold is marked as a low stress response region. The second threshold is taken as the 25th percentile value of the stress amplitude in the current frame, used to identify abnormal areas of stress release and surrounding rock loosening. Using isosurfaces to delineate boundaries can accurately match the actual distribution of the stress field. Compared with the fixed radius method, it is more in line with the physical laws of stress diffusion in the surrounding rock, and the delineated influence range is more in line with the actual engineering situation.

[0040] Finally, all identified high-stress response regions and low-stress response regions within the current frame are integrated, and each region is assigned a unique region number, spatial coordinate range, stress amplitude interval, and corresponding time label to form a structured anchor stress response zoning information output. The zoning results can be directly mapped to the engineering BIM model for visualization, providing an intuitive spatial situation basis for subsequent anomaly type determination.

[0041] The above steps achieve robust extraction of stress characteristics through sliding window time-series statistics, complete the accurate mapping from discrete measurement points to a continuous three-dimensional stress field through Kriging spatial interpolation, and achieve automatic identification of stress anomaly regions through extreme value search and isosurface extraction. The entire processing chain is progressive from time-series to space, from features to field, and from identification to partitioning. Each link provides a feasible engineering implementation method, which not only overcomes the limitation that single-point monitoring cannot reflect the stress situation of the whole domain, but also ensures the physical rationality and computational feasibility of spatial interpolation and partitioning identification. It can provide high-precision whole-domain spatial data support for subsequent stress-displacement correlation analysis and graded early warning judgment, effectively improving the spatial resolution and engineering practical value of anchor bolt stress monitoring and analysis.

[0042] For step S3, the stress time-series monitoring data and the displacement time-series monitoring data are aligned by timestamp, and the correlation degree between the stress measurement and displacement measurement at the same time is calculated to obtain the correlation feature parameter. This includes: at the same time, pairing the stress measurement and displacement measurement that match in spatial location into monitoring data pairs; performing normalization processing on the stress measurement and displacement measurement in the monitoring data pairs respectively; calculating the Pearson correlation coefficient of the normalized monitoring data pairs, and using the Pearson correlation coefficient as the correlation feature parameter.

[0043] Specifically, the first step was to construct spatiotemporally matched monitoring data pairs. In the temporal dimension, using the acquisition timestamps of the preprocessed stress time-series monitoring data as a benchmark, the displacement time-series monitoring data was resampled using linear interpolation. This precisely matched the displacement measurements to each stress acquisition moment, eliminating time misalignment caused by sampling clock deviations and sampling frequency differences between the two types of sensors. This ensured that the stress and displacement measurements of the same data pair strictly corresponded to the same physical moment, providing an accurate time reference for subsequent correlation calculations. In the spatial dimension, based on the installation section number, surrounding rock depth, and three-dimensional coordinates recorded in the on-site monitoring equipment log, anchor stress gauges and multi-point displacement gauges located at the same monitoring section and burial depth were paired one-to-one. This ensured that each monitoring data pair originated from the mechanical response of the same surrounding rock area, avoiding distortion in correlation calculations due to spatial misalignment and guaranteeing the physical meaning of the correlation analysis. After completing the spatiotemporal matching, several corresponding stress-displacement time-series data pairs were formed. Each data pair contained a stress measurement sequence and a displacement measurement sequence of equal length.

[0044] To eliminate the interference of dimensional and numerical differences between stress and displacement on the correlation coefficient calculation results, min-max normalization was performed on both sets of sequences before calculating the correlation coefficient, mapping the measured values ​​uniformly to the [0,1] interval. The normalization formula for the stress sequence is as follows: , The normalization formula for the displacement sequence is: , Where, σ k δ k These are the original measured values ​​of stress and displacement at time k, respectively, σ max σ min δ represents the maximum and minimum values ​​of the stress sequence within the current analysis window. max δ min Let σ' be the maximum and minimum values ​​of the displacement sequence. k ,δ' kThe values ​​are normalized and range from [0,1]. By normalizing, the changing trends of two different physical quantities can be compared on a unified numerical scale, avoiding the dominance of the larger numerical value in the correlation coefficient due to the large difference in numerical magnitude. This ensures that the correlation calculation can truly reflect the degree of synergy between the changing trends of the two quantities, rather than the degree of matching of numerical magnitudes.

[0045] After normalization, the Pearson correlation coefficient r between the two sets of sequences is calculated and used as the association feature parameter. The calculation formula is as follows: , Where m represents the total number of valid data pairs in the current analysis window, and the analysis window has the same length as the second preset time window, updating synchronously. The mean of the normalized stress sequence, The mean of the normalized displacement sequence is given. The Pearson correlation coefficient r ranges from -1 to 1. The closer its absolute value is to 1, the stronger the correlation between stress and displacement trends; the closer its absolute value is to 0, the weaker the correlation between their trends. In engineering applications, a judgment threshold is usually set. When |r| ≥ 0.7, a significant synergistic relationship between stress and displacement is considered, and subsequent anomaly judgments tend to favor the overall deformation of the surrounding rock type. When |r| < 0.7, no significant synergistic relationship is considered, and subsequent anomaly judgments tend to favor the local factor type. Quantifying the correlation between stress and displacement using the Pearson correlation coefficient can transform qualitative synergistic judgments into quantitative numerical indicators, providing a reusable and quantifiable basis for anomaly type judgment in subsequent early warning stages.

[0046] The above steps construct monitoring data pairs with clear physical correspondences through time interpolation alignment and spatial location pairing, ensuring the rationality of the correlation analysis from the data source. Normalization eliminates calculation biases caused by dimensional differences, ensuring that the correlation results objectively reflect trend synergy. Finally, the correlation characteristic parameters are quantitatively output through the Pearson correlation coefficient, achieving accurate measurement of the stress-displacement synergy. This provides core discrimination indicators for classifying different anomaly types such as overall rock deformation, local geological influence, and local disturbance in subsequent graded early warning, effectively making up for the inability of a single stress indicator to classify and characterize anomalies, and significantly improving the engineering guidance value of the early warning results.

[0047] For step S4, based on the changing trend of the stress time-series monitoring data within the second preset time window, a warning judgment threshold is dynamically generated. If the real-time stress measurement value exceeds the warning judgment threshold, a comprehensive judgment is made by combining the anchor stress response zoning information and the associated feature parameters to generate a graded warning signal.

[0048] The step of dynamically generating an early warning threshold based on the changing trend of the stress time-series monitoring data within a second preset time window includes: acquiring stress measurements at each acquisition time within the second preset time window; calculating the stress change rate based on the stress measurements at the start and end times of the second preset time window, and calculating the stress fluctuation amplitude based on the stress measurements at each acquisition time within the second preset time window; if the stress change rate does not exceed a preset change rate threshold, then linearly adjusting the initial stress value based on the stress fluctuation amplitude to generate an early warning threshold; if the stress change rate exceeds the preset change rate threshold, then nonlinearly adjusting the initial stress value based on the stress change rate and the stress fluctuation amplitude to generate an early warning threshold.

[0049] If the real-time stress measurement exceeds the warning threshold, a comprehensive judgment is made based on the anchor stress response zoning information and the associated feature parameters to generate a graded warning signal. This includes: when the real-time stress measurement exceeds the warning threshold, marking the spatial coordinates of the anchor stress gauge corresponding to the real-time stress measurement as an abnormal measurement point location; obtaining the anchor stress response zoning attribution and corresponding associated feature parameters of the abnormal measurement point location; if the abnormal measurement point location is located within the high-stress response area and the associated feature parameters indicate a cooperative change relationship, then a general deformation anomaly of the surrounding rock is determined, and a Level 1 warning signal is generated; if the abnormal measurement point location is located within the high-stress response area and the associated feature parameters indicate no cooperative change relationship, then a local geological impact anomaly is determined, and a Level 2 warning signal is generated; if the abnormal measurement point location is not located within the high-stress response area and the associated feature parameters indicate a cooperative change relationship, then an adjacent area deformation impact anomaly is determined, and a Level 2 warning signal is generated; if the abnormal measurement point location is not located within the high-stress response area and the associated feature parameters indicate no cooperative change relationship, then a local disturbance anomaly is determined, and a Level 3 warning signal is generated.

[0050] Specifically, the dynamic early warning threshold is first calculated using a rolling calculation. The length of the second preset time window is set to T2, which can be configured within the range of 24h ≤ T2 ≤ 168h based on the surrounding rock deformation rate and engineering risk level. In this embodiment, T2 = 72h is used. The window is updated periodically with the monitoring data, and the threshold parameter is recalculated each control cycle to ensure that the early warning threshold always matches the latest anchor stress evolution trend. Effective stress measurements at all acquisition times within the current rolling window are extracted, and two core trend indicators, stress change rate and stress fluctuation amplitude, are calculated. The stress change rate k reflects the overall rate of stress rise and fall within the window, and the calculation formula is: , Where, σ start The stress measurement is σ at the start of the second time window.end The stress measurement at the end of the window is used to calculate the stress change rate k, which is in MPa / h. A positive value indicates an increasing stress trend, while a negative value indicates a decreasing stress trend. This index can quantitatively characterize the overall direction and severity of recent stress changes and is a core basis for classifying stress evolution conditions and selecting threshold adjustment modes. Stress fluctuation amplitude σ amp The formula for calculating the overall fluctuation range of stress within the window is as follows: , Where, σ max,T2 σ min,T2 These represent the maximum and minimum stress values ​​measured within the second time window. A larger fluctuation indicates weaker stability of the anchor bolt stress state and a wider range of normal fluctuations.

[0051] Preset rate of change threshold k th As the boundary between the two threshold adjustment modes, this embodiment takes k. th =0.05MPa / h, this value can be adjusted within the range of 0.02MPa / h to 0.1MPa / h according to engineering geological conditions and risk tolerance. When the absolute value of the stress change rate does not exceed the preset change rate threshold, i.e., |k|≤k th When the condition is determined to be a gradually changing working condition, the overall stress is in a relatively stable state. The initial stress value is linearly adjusted only based on the stress fluctuation amplitude to generate the early warning judgment threshold σ. warn The calculation formula is: , Where, σ base The initial stress value is determined during the engineering design phase based on the anchor bolt's design bearing capacity and surrounding rock strength parameters. It is typically taken as 60% to 70% of the anchor bolt's yield stress and serves as the basic benchmark for early warning judgment. α is the linear adjustment coefficient, ranging from 0.1 to 0.3; in this embodiment, it is set to 0.2. The linear adjustment mode under smooth working conditions can synchronously raise the early warning threshold with the normal fluctuation range of stress, avoiding frequent early warnings triggered by normal small fluctuations in anchor bolt stress, effectively reducing the false alarm rate and improving the stability of the early warning results.

[0052] When the rate of change of stress exceeds the preset rate of change threshold, i.e., k>k th When the stress shows a rapid upward trend, it is judged as a rapidly changing working condition. At this time, the stress in the anchor bolt rises rapidly, and the risk of instability increases significantly. It is necessary to combine the stress change rate and stress fluctuation amplitude to make nonlinear adjustments to the initial stress value to generate an early warning judgment threshold σ. warn The calculation formula is: , Where β is the rate of change influence coefficient, ranging from 0.2 to 0.5, and 0.3 in this embodiment, it is used to control the tightening effect of the stress rise rate on the threshold. γ is the fluctuation amplitude influence coefficient, ranging from 0.15 to 0.4, and 0.2 in this embodiment. In this formula, the faster the stress rise rate, the greater the reduction of the first term, and the lower the final warning threshold. σ is also specified during the calculation process. warn ≥0.6σ base To avoid excessively lowering the threshold and losing the significance of the early warning benchmark, the system can automatically tighten the early warning standard during the rapid stress rise phase, achieving early warning of risks and effectively compensating for the shortcomings of fixed thresholds in responding to sudden stress increases, thus improving the early warning sensitivity to rapid instability risks. If the stress shows a rapid downward trend, the threshold is adjusted only according to the linear mode of the smooth working condition, avoiding unnecessary early warning triggering during the stress decline phase.

[0053] After completing the real-time calculation of the dynamic early warning threshold, the real-time stress measurement value of each measuring point is compared with the current early warning judgment threshold one by one. Once the real-time stress measurement value of a measuring point exceeds the corresponding threshold, the anomaly judgment process is immediately triggered, and the multi-factor comprehensive hierarchical early warning logic is initiated. First, based on the anchor stress gauge device number that triggered the anomaly, its installation three-dimensional coordinates are retrieved from the spatial database and marked as the location of the abnormal measuring point. Then, two types of auxiliary judgment information are retrieved simultaneously: one is the anchor stress response zoning information of the current time frame, which determines whether the abnormal measuring point falls within the identified high stress response area through spatial coordinate matching; the other is the correlation characteristic parameter calculated from the displacement monitoring data of the same location corresponding to the measuring point, namely the Pearson correlation coefficient, to determine whether there is a significant synergistic change relationship between stress and displacement.

[0054] Combining the four scenarios from the two auxiliary judgment dimensions, corresponding to different anomaly causes and risk levels, corresponding warning signals are generated: When the anomaly measuring point is located within a high-stress response area, and the associated characteristic parameter satisfies |r|≥0.7 and there is a synergistic change relationship, it indicates that the stress anomaly develops synchronously with the overall deformation of the surrounding rock, and the high-stress response continues to intensify. This is the case with the highest risk of large-scale surrounding rock instability, and is judged as an anomaly of overall surrounding rock deformation, generating a Level 1 warning signal. This level of warning corresponds to the highest priority of handling, requiring immediate on-site verification and the initiation of prevention and control measures. When the anomaly measuring point is located within a high-stress response area, but the associated characteristic parameter satisfies |r|<0.7 and there is no significant synergistic change relationship, it indicates that the stress increase is not accompanied by a large-scale displacement increase. The anomaly is not dominated by overall surrounding rock deformation, but is more likely caused by local geological structural factors such as local joint surface slippage and tectonic stress release. The risk level is secondary, and it is judged as an anomaly of local geological influence, generating a Level 2 warning signal. It is necessary to focus on the stability evolution of the local rock mass. When the anomalous measuring point is not located within the high-stress response area, but the associated characteristic parameter satisfies |r|≥0.7 and there is a synergistic change relationship, it indicates that the stress anomaly is the result of the overall deformation transmission of the surrounding rock in the surrounding area. The anomaly has diffusion and transmission properties. Although the current stress level has not reached the intensity of the concentration area, there is still a possibility of subsequent risk escalation. It is judged as an anomaly influenced by deformation in the adjacent area, and a level-two warning signal is generated. It is necessary to continuously track the deformation diffusion trend. When the anomalous measuring point is not located within the high-stress response area, and the associated characteristic parameter satisfies |r|<0.7 and there is no synergistic change relationship, it indicates that the stress anomaly is only a single-point local phenomenon without overall deformation support. It is most likely caused by instantaneous and local factors such as on-site blasting disturbance, instantaneous closure of joints and fissures, or slight sensor drift. The overall risk is the lowest. It is judged as a local disturbance anomaly, and a level-three warning signal is generated. Only continuous monitoring and observation are required, and no emergency response is needed.

[0055] The above steps, through a trend-adaptive dynamic early warning threshold mechanism, overcome the limitation that fixed thresholds cannot adapt to different stress evolution conditions. This controls the false alarm rate under stable working conditions and improves early warning sensitivity under rapidly changing conditions, balancing accuracy and timeliness. Simultaneously, a three-level progressive judgment logic is constructed, consisting of threshold triggering, spatial partition verification, and displacement correlation verification. This not only outputs the early warning level but also simultaneously clarifies the causes of anomalies and the sources of risks, making the early warning results more meaningful for engineering guidance. This effectively solves the shortcomings of traditional anchor bolt stress early warning systems, which can only provide over-limit warnings and cannot distinguish between risk types and causes, significantly improving the reliability and practical engineering value of the early warning results.

[0056] For step S5, based on the graded early warning signal, output the anchor bolt stress analysis results, which include the location of the early warning measuring point and the early warning level.

[0057] The step of outputting anchor bolt stress analysis results containing the location of the warning measurement point and the warning level based on the graded warning signal includes: parsing the graded warning signal, extracting the warning level identifier carried by the graded warning signal and the device number of the anchor bolt stress gauge that issued the graded warning signal; retrieving the pre-stored device ledger based on the device number, and retrieving the installation coordinates of the anchor bolt stress gauge corresponding to the device number as the location of the warning measurement point; converting the warning level identifier into a corresponding warning level text description; and writing the warning measurement point location and the warning level text description into the corresponding fields of the analysis result record table to generate the anchor bolt stress analysis results.

[0058] After generating the anchor bolt stress analysis results, the process further includes: defining a verification range centered on the location of the warning measurement point with a warning level of Level 1 in the anchor bolt stress analysis results; extracting the stress time-series monitoring data of all anchor bolt stress gauges within the verification range in the second preset time window; sequentially performing anchor bolt stress response field construction, anchor bolt stress response partition identification, related feature parameter calculation, and graded warning signal generation on the stress time-series monitoring data to obtain the verification judgment results; and writing the verification judgment results into the anchor bolt stress analysis results.

[0059] Specifically, the generated graded early warning signals are first subjected to structured parsing. Each graded early warning signal is a standardized structured data packet, containing fields such as the unique device number of the anchor stress gauge that triggered the early warning, the early warning level enumeration identifier, the early warning trigger timestamp, the real-time stress measurement value at the trigger time, the corresponding associated characteristic parameter values, and the preliminary anomaly type judgment code. The parsing process extracts core information bit by bit according to the preset field definitions, avoiding data misalignment and information omission, and ensuring the accuracy and completeness of the early warning information extraction from the output end. After the field extraction is completed, the monitoring equipment ledger, which is pre-stored in the system database, is queried using the device number as the search key. This ledger is established synchronously during the sensor installation phase, recording each device's unique number, installation section pile number, three-dimensional spatial coordinates, surrounding rock burial depth, sensor model, installation date, and calibration record, among other full attribute information. The three-dimensional coordinates are entered after high-precision measurement using a total station during the construction phase, with coordinate errors controlled at the centimeter level. The installation three-dimensional coordinates and section information of the corresponding device are retrieved as the early warning measurement point location, ensuring that the spatial accuracy of the early warning location meets the actual needs of on-site positioning verification.

[0060] Next, the system performs text mapping conversion of the warning levels. It incorporates unified level mapping rules to convert numerical warning level enumeration identifiers into standardized engineering text descriptions. Enumeration value 1 corresponds to a Level 1 warning, value 2 to a Level 2 warning, and value 3 to a Level 3 warning. Simultaneously, it combines the anomaly type determination code in the warning signal to supplement corresponding anomaly type description text, such as: overall rock deformation type, local geological impact type, local disturbance type, etc. This ensures the output not only includes level information but also a clear causal indication, facilitating direct understanding of the warning meaning by engineers without secondary interpretation, thus improving readability and engineering practicality. After all information extraction and conversion are completed, the 3D coordinates of the warning measurement points, cross-sectional location, warning level, causal description, trigger time, real-time stress value, and associated characteristic parameters are written into the corresponding fields of the preset analysis result record table, forming a single complete warning analysis record. Summarizing all warning records within the current period generates a structured anchor stress analysis result. This result can be directly exported as a standardized engineering report or simultaneously pushed to the engineering safety monitoring platform for visualization, enabling rapid flow and intuitive presentation of warning information.

[0061] For the highest-risk Level 1 warning, a verification mechanism is automatically initiated after generating preliminary analysis results to avoid high-risk false alarms caused by factors such as single-point sensor failures and instantaneous data jumps, ensuring the seriousness and reliability of the Level 1 warning. First, a spherical space with a radius of 5m is defined as the verification range, centered on the three-dimensional coordinates of the Level 1 warning measurement point. This range covers all adjacent anchor stress gauges around the measurement point, ensuring the spatial representativeness of the verification sample without introducing data from irrelevant areas that could interfere with the verification judgment due to an excessively large range. The verification range can be flexibly adjusted within the range of 3m to 8m based on the measurement point density and the engineering risk level. Then, the full stress time-series monitoring data of all valid anchor stress gauges within the verification range are extracted within a second preset time window. The complete processing chain of the aforementioned steps is invoked, sequentially executing stress statistical feature extraction, Kriging interpolation to construct the local anchor stress response field, identification of local high-stress and low-stress response regions, and calculation of stress-displacement correlation characteristic parameters for each measurement point. Finally, the graded warning signals for all measurement points within this local area are regenerated, yielding the verification judgment result.

[0062] The review process distinguishes between single-point anomalies and regional risks through cross-validation of multiple neighboring points. If more than 30% (30% is an engineering experience threshold that can be adjusted based on the engineering risk level and the density of measuring points) of the measuring points within the review scope simultaneously show a Level II or higher warning level, and a continuous high-stress response area is identified in the local stress field, then the Level I warning is confirmed as valid, and the area is judged as a regional risk of surrounding rock instability, requiring immediate on-site verification and emergency response. If only the single point triggering the warning has a stress anomaly within the review scope, while the stress state of surrounding measuring points is at a normal level and there are no continuous high-stress response characteristics, then the anomaly is judged as a single-point false alarm caused by sensor drift or a localized minute disturbance. The warning level is downgraded to Level III and marked as a single-point anomaly, and it is recommended to check the sensor status. Finally, the review conclusion, review coverage, number of measuring points participating in the review, and final level after review are added to the review field of the corresponding warning record to form a complete anchor bolt stress analysis result containing the preliminary judgment and review conclusion.

[0063] The above steps, through standardized signal analysis, ledger retrieval, and structured record generation, achieve a standardized transformation of early warning information from internally calculated signals to usable engineering results. The output content is complete, accurately located, and clearly categorized, directly supporting on-site safety assessment and response decisions. The accompanying automatic verification mechanism for Level 1 early warnings effectively filters high-risk false alarms without increasing manual costs through multi-point cross-verification in neighboring regions. This avoids unnecessary emergency response costs caused by false Level 1 early warnings and prevents the omission of real regional risks, significantly improving the credibility and engineering guidance value of the highest-level early warnings. This makes the output results of the entire analysis method more rigorous and more in line with the actual application needs of engineering projects.

[0064] In summary, the anchor stress analysis method based on monitoring time-series data provided by this invention extends discrete single-point monitoring into a continuous anchor stress response field through time-series statistics and Kriging space interpolation, breaking through the spatial limitations of traditional point monitoring. The introduction of stress-displacement correlation quantitative analysis allows for the classification of multiple anomaly types, compensating for the inability of a single stress index to qualitatively determine the source of risk. Combining trend-adaptive dynamic thresholds and multi-factor hierarchical judgment, it simultaneously considers both early warning sensitivity and accuracy. The accompanying first-level early warning verification mechanism further ensures the reliability of high-risk early warnings. The entire solution effectively improves the intelligence level and early warning accuracy of anchor stress monitoring in underground engineering, providing reliable decision support for surrounding rock stability evaluation and geological disaster prevention.

[0065] like Figure 2 The system shown is the one used for the anchor bolt stress method based on monitoring time-series data in this application, and includes the following modules: The data acquisition and preprocessing module is used to acquire stress time-series monitoring data from anchor stress gauges and displacement time-series monitoring data from multi-point position gauges on the same cross section, and to remove invalid data from both types of data. The anchor stress field and zoning module is used to extract the stress statistical characteristics of the stress time series monitoring data according to the first preset time window, interpolate according to the spatial distribution of the anchor stress gauge, construct the anchor stress response field, identify the high stress response area and low stress response area according to the anchor stress response field, and output the anchor stress response zoning information. The stress-displacement correlation calculation module is used to align the stress time-series monitoring data and the displacement time-series monitoring data by timestamp, calculate the correlation between the stress measurement value and the displacement measurement value at the same moment, and obtain the correlation characteristic parameters. The dynamic threshold and graded early warning module is used to dynamically generate an early warning judgment threshold based on the changing trend of the stress time series monitoring data within a second preset time window. If the real-time stress measurement value exceeds the early warning judgment threshold, a comprehensive judgment is made by combining the anchor stress response zoning information and the associated feature parameters to generate a graded early warning signal. The analysis result output module is used to output anchor bolt stress analysis results, including the location of the warning measuring point and the warning level, based on the graded warning signal.

[0066] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any way, and all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A method for analyzing anchor bolt stress based on monitoring time-series data, characterized in that, Includes the following steps: S1. Collect stress time-series monitoring data from anchor stress gauges and displacement time-series monitoring data from multi-point displacement gauges on the same cross section, and remove invalid data from both types of data; S2. Extract the stress statistical characteristics of the stress time series monitoring data according to the first preset time window, interpolate according to the spatial distribution of the anchor stress gauge, construct the anchor stress response field, identify the high stress response area and low stress response area according to the anchor stress response field, and output the anchor stress response zoning information. S3. Align the stress time-series monitoring data and the displacement time-series monitoring data according to the timestamp, calculate the correlation between the stress measurement value and the displacement measurement value at the same moment, and obtain the correlation characteristic parameters; S4. Based on the changing trend of the stress time-series monitoring data within the second preset time window, a warning judgment threshold is dynamically generated. If the real-time stress measurement value exceeds the warning judgment threshold, a comprehensive judgment is made by combining the anchor stress response zoning information and the associated feature parameters to generate a graded warning signal. S5. Based on the graded early warning signal, output the anchor bolt stress analysis results, including the location of the early warning measuring point and the early warning level.

2. The anchor bolt stress analysis method based on monitoring time series data according to claim 1, characterized in that, The extraction of stress statistical features from the stress time-series monitoring data according to the first preset time window includes: Set the window length and sliding step size of the first preset time window, and sequentially truncate the stress time series monitoring data according to the sliding step size based on the window length to obtain multiple stress time series segments; The stress timing segments are sequentially continuous and adjacent segments partially overlap. The arithmetic mean, maximum, minimum, median, and standard deviation of the stress measurements within the stress time series segment are calculated and used as the stress statistical feature vector of the stress time series segment.

3. The anchor bolt stress analysis method based on monitoring time series data according to claim 2, characterized in that, The step of interpolating based on the spatial distribution of the anchor stress gauges to construct the anchor stress response field includes: Using the three-dimensional coordinates of the anchor stress gauge in the spatial coordinate system of the cavern group as interpolation sample points, and the stress statistical feature vector corresponding to the first preset time window of the interpolation sample points as interpolation attributes, an interpolation grid is laid out within the spatial range of the surrounding rock of the cavern group. The stress statistical eigenvectors at the grid nodes are estimated unbiased optimally using the Kriging interpolation method. The average value of the stress statistical feature vector at the grid node is extracted as the stress amplitude. Combined with the grid node coordinates and the first preset time window index, the anchor stress response field is generated.

4. The anchor bolt stress analysis method based on monitoring time series data according to claim 3, characterized in that, The method of using Kriging interpolation to perform unbiased optimal estimation of the stress statistical eigenvectors at grid nodes includes: The spatial distance between any two points is calculated based on the three-dimensional coordinates of the interpolated sample points. The attribute difference is calculated for each statistical component of the feature vector. A semi-variogram model is fitted based on the correspondence between the spatial distance and the attribute difference. The Kriging weight coefficients between the grid nodes and the interpolated sample points are determined based on the semi-variogram model. Based on the Kriging weight coefficients, the stress statistical feature vectors of the interpolated sample points are weighted and summed to obtain the stress statistical feature vectors of the grid nodes.

5. The anchor bolt stress analysis method based on monitoring time series data according to claim 1, characterized in that, The step of identifying high-stress response regions and low-stress response regions based on the anchor bolt stress response field and outputting anchor bolt stress response zoning information includes: Locate the local maxima and local minima of stress amplitude at each spatial location in the anchor bolt stress response field; Using the local maximum point as the center and the stress amplitude isosurface at that point as the boundary, the high stress response influence range is defined, and the region within the high stress response influence range where the stress amplitude exceeds a preset first threshold is marked as the high stress response region. Using the local minimum point as the center and the stress amplitude isosurface at that point as the boundary, the low stress response influence range is defined, and the region within the low stress response influence range where the stress amplitude is lower than a preset second threshold is marked as the low stress response region. All marked high-stress response areas and low-stress response areas are integrated into anchor bolt stress response zoning information.

6. The anchor bolt stress analysis method based on monitoring time series data according to claim 1, characterized in that, The correlation between stress and displacement measurements at the same time is calculated to obtain correlation characteristic parameters, including: At the same time, stress measurements and displacement measurements that match the spatial location are paired as monitoring data pairs; Normalization is performed on the stress and displacement measurements in the monitoring data pairs respectively; Calculate the Pearson correlation coefficient of the normalized monitoring data pairs, and use the Pearson correlation coefficient as the correlation feature parameter.

7. The anchor bolt stress analysis method based on monitoring time series data according to claim 1, characterized in that, The step of dynamically generating an early warning threshold based on the changing trend of the stress time-series monitoring data within a second preset time window includes: Obtain the stress measurement values ​​at each acquisition time within the second preset time window; The stress change rate is calculated based on the stress measurements at the start and end times of the second preset time window, and the stress fluctuation amplitude is calculated based on the stress measurements at each acquisition time within the second preset time window. If the stress change rate does not exceed the preset change rate threshold, the initial stress value is linearly adjusted based on the stress fluctuation amplitude to generate an early warning judgment threshold. If the stress change rate exceeds the preset change rate threshold, the initial stress value is nonlinearly adjusted based on the stress change rate and the stress fluctuation amplitude to generate a warning judgment threshold.

8. The anchor bolt stress analysis method based on monitoring time series data according to claim 7, characterized in that, If the real-time stress measurement value exceeds the warning threshold, a comprehensive judgment is made based on the anchor bolt stress response zoning information and the associated feature parameters to generate a graded warning signal, including: When the real-time stress measurement value exceeds the warning judgment threshold, the spatial coordinates of the anchor stress gauge corresponding to the real-time stress measurement value are marked as the abnormal measurement point location. Obtain the anchor stress response zoning at the abnormal measurement point location and the corresponding associated characteristic parameters; If the location of the abnormal measuring point is within the high-stress response area and the associated characteristic parameters indicate a cooperative change relationship, then the overall deformation of the surrounding rock is determined to be abnormal, and a first-level early warning signal is generated. If the location of the abnormal measuring point is within the high-stress response area and the associated characteristic parameters indicate that there is no coordinated change relationship, then a local geological impact anomaly is determined, and a secondary early warning signal is generated. If the location of the abnormal measuring point is not located within the high stress response area and the associated characteristic parameters indicate a cooperative change relationship, then the anomaly is determined to be a deformation-affected anomaly in the adjacent area, and a secondary early warning signal is generated. If the location of the abnormal measuring point is not within the high-stress response area and the associated characteristic parameters indicate no coordinated change relationship, then a local disturbance anomaly is determined, and a level three warning signal is generated.

9. The anchor bolt stress analysis method based on monitoring time series data according to claim 1, characterized in that, The anchor bolt stress analysis results, including the location of the warning measurement point and the warning level, are output based on the graded warning signal, including: The graded early warning signal is analyzed to extract the early warning level identifier and the device number of the anchor stress gauge that issued the graded early warning signal. Based on the equipment number, retrieve the pre-stored equipment ledger and retrieve the installation coordinates of the anchor stress gauge corresponding to the equipment number as the location of the early warning measurement point; Convert the warning level identifier into a corresponding warning level text description; Write the location of the early warning measurement point and the text description of the early warning level into the corresponding fields of the analysis result record table to generate the anchor bolt stress analysis results.

10. The anchor bolt stress analysis method based on monitoring time series data according to claim 9, characterized in that, After generating the anchor bolt stress analysis results, the following is also included: The verification range is defined centered on the location of the warning measurement point with a warning level of Level 1 in the anchor bolt stress analysis results; Extract stress time-series monitoring data of all anchor stress gauges within the second preset time window within the scope of the review, and sequentially perform anchor stress response field construction, anchor stress response partition identification, related feature parameter calculation and graded early warning signal generation on the stress time-series monitoring data to obtain the review judgment result; The verification result is written into the anchor bolt stress analysis result.