Surrounding rock failure early warning method based on multi-source data

Abstract

The invention discloses a surrounding rock failure early warning method based on multi-source data, and belongs to the technical field of data processing, and the method comprises the steps: arranging a plurality of observation stations on a working face along a withdrawing channel according to a preset interval, and installing sensor groups on a top plate, a real coal side and a roadside filling body of each observation station; carrying out data processing on the collected data by adopting a moving average method, fast Fourier transform, difference and cumulative summation, and aligning the processed data according to the measuring point position and the time label; performing principal component analysis and hierarchical clustering on the aligned data, dividing a measurement point into three regions, respectively establishing corresponding prediction models, and performing normalization processing on model calculation results; and setting an abnormity judgment parameter according to a normalization processing result, calculating an early warning index by adopting weighted summation, and sending out an early warning signal according to a comparison result of the early warning index and a preset grade threshold value. Through the scheme of the invention, the deformation trend of the surrounding rock in each region can be predicted more accurately.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a surrounding rock damage early warning method based on multi-source data. Background Art

[0002] With the continuous expansion of coal mining scale and the continuous deepening of mining depth, the dynamic disasters caused by the destruction of coal mine surrounding rocks are becoming more and more serious, posing a major threat to the safe production of coal mines. Coal mine surrounding rock destruction early warning technology refers to a technical method that uses a variety of monitoring methods to obtain the changing characteristics of surrounding rock deformation, stress and other parameters, and combines intelligent analysis algorithms to predict and warn the occurrence of surrounding rock destruction in advance. The core of this technology is to timely discover the precursors of surrounding rock destruction by real-time monitoring of the surrounding rock stability state, and disclose the decision-making basis for taking preventive and control measures. The existing coal mine surrounding rock destruction early warning technology mainly includes displacement monitoring, stress monitoring, microseismic monitoring and other methods. Among them, displacement monitoring technology determines the stability of the surrounding rock by measuring the deformation of the tunnel roof and wall; stress monitoring technology assesses the risk of destruction by measuring the changes in the surrounding rock stress field; microseismic monitoring technology predicts destruction by capturing the weak seismic signals generated during the surrounding rock rupture process. However, the current conventional early warning technology has many shortcomings: first, a single monitoring method is difficult to fully reflect the complex evolution process of surrounding rock failure, resulting in low reliability of early warning results; second, traditional early warning methods often rely on experience and judgment, lacking a scientific early warning indicator system and early warning model; third, due to the complexity and variability of coal mine geological conditions, the adaptability and accuracy of existing early warning technologies under different geological conditions vary greatly; finally, the deployment and maintenance costs of the early warning system are high, and the monitoring equipment is prone to malfunction in harsh underground environments, affecting the continuity and reliability of monitoring data.

[0003] In response to the above technical problems, researchers have disclosed a variety of improvement plans. The first is to develop multi-parameter collaborative monitoring technology, and establish a more comprehensive early warning indicator system by integrating multiple monitoring data such as displacement, stress, and microseismicity. The second is to introduce artificial intelligence algorithms, such as deep learning and pattern recognition, to improve the intelligence level and prediction accuracy of the early warning model. The third is to develop an adaptive early warning model to improve the applicability of early warning technology under different conditions by considering influencing factors such as geological conditions and mining methods. In addition, it also includes the development of new sensors and monitoring equipment to improve the reliability and durability of monitoring equipment. However, these improved technologies still have some problems in practical applications: first, the fusion method of multi-parameter monitoring data is not yet mature, and it is difficult to effectively extract and utilize the early warning information contained in various types of data; second, the stability and interpretability of artificial intelligence algorithms in dealing with nonlinear and non-stationary surrounding rock failure processes need to be improved; third, due to the lack of long-term and systematic field monitoring data, the training and verification of early warning models are difficult; finally, the intelligence level and anti-interference ability of monitoring equipment still need to be further improved to adapt to the complex underground environment. In addition, the real-time and reliability of the early warning system also face challenges. How to shorten the warning time while ensuring the accuracy of the warning, and how to reduce the false alarm rate and missed alarm rate are technical problems that need to be solved urgently.

[0004] Therefore, a technical solution is urgently needed to more accurately predict the deformation trend of the surrounding rock in each region. Summary of the invention

[0005] In order to solve the shortcomings of the prior art, the embodiment of the present application discloses a surrounding rock damage early warning method based on multi-source data. The present application solves the technical problems of the prior art such as single model and linear superposition.

[0006] The embodiment of the present application discloses a surrounding rock damage early warning method based on multi-source data, including: setting up multiple measuring stations at predetermined intervals along the withdrawal channel of the working face, and installing sensor groups on the roof, solid coal wall, and filling body beside the roadway at the measuring stations; using the sliding average method, fast Fourier transform, difference and cumulative summation to process the collected data, and aligning the processed data according to the measuring point position and time label; performing principal component analysis and hierarchical clustering on the aligned data, dividing the measuring points into three areas, establishing corresponding prediction models respectively, and normalizing the model calculation results; setting abnormal judgment parameters according to the normalization processing results, using weighted summation to calculate the early warning index, and issuing an early warning signal based on the comparison result of the early warning index with a preset level threshold.

[0007] In a possible implementation, a plurality of measuring stations are set at predetermined intervals along the withdrawal channel of the working face, and a sensor group is installed on the roof, solid coal wall, and filling body beside the roadway of the measuring stations, including: a plurality of measuring stations are arranged in sequence at preset intervals along the withdrawal channel of the working face, and a pressure sensor, an axial stress sensor, a displacement sensor, and a speed sensor are installed at each measuring station; pressure sensors are respectively installed on the roof, solid coal wall, and filling body beside the roadway of the working face, and an axial stress sensor is installed at the end measuring station; the measurement data of the sensors are collected and stored by a data collector to form a data file containing pressure, stress, displacement, and deformation speed.

[0008] In a possible implementation, the collected data is processed by using a sliding average method, a fast Fourier transform, a difference and a cumulative summation, and the processed data is aligned according to the measuring point position and the time label, including: the collected pressure data is smoothed by using a sliding average method, and the spatial distribution characteristics of the pressure data are fitted by the least squares method; the frequency spectrum characteristics of the anchor axial stress data are calculated by the fast Fourier transform method and the main frequency components are extracted; the displacement data is calculated by the difference method The displacement increment is calculated and the time series characteristics are counted; the deformation velocity data is calculated by the cumulative summation method. The accumulated deformation amount; the processed data is aligned according to the measuring point position and the time label to form a processed data table.

[0009] In a possible implementation, principal component analysis and hierarchical clustering are performed on the aligned data, the measuring points are divided into three areas, corresponding prediction models are established for each area, and the model calculation results are normalized, including: using principal component analysis to extract the main change characteristics of the data of each measuring point, and calculating the correlation coefficient between each measuring point; using hierarchical clustering to divide the measuring points into three areas behind the stop mining line, above the withdrawal channel, and above the coal wall of the working face; establishing a prediction equation of the creep constitutive model for the area behind the stop mining line; establishing a prediction equation of the viscoelastic-plastic constitutive model for the area above the withdrawal channel; establishing a prediction equation of the rheological damage coupling model for the area above the coal wall; and normalizing the predicted values ​​of the three areas to obtain normalized predicted values.

[0010] In a possible implementation, abnormal judgment parameters are set according to the normalization processing results, the warning index is calculated by weighted summation, and a warning signal is issued based on the comparison result of the warning index and the preset level threshold, including: setting judgment parameters for pressure abnormality, stress abnormality, displacement abnormality and velocity abnormality; using the weighted summation method to calculate the warning index of each measuring point, and the weight coefficient is determined by experimental calibration; the warning index is divided into preset levels according to the numerical value, and a warning signal is issued when the warning index exceeds the corresponding level threshold.

[0011] In a possible implementation, principal component analysis is used to extract the main change characteristics of the data of each measuring point, and the correlation coefficient between each measuring point is calculated, including: subtracting the measurement mean of the top plate pressure, anchor load, top and bottom plate movement and two-side movement data of each measuring point for a preset time period before the measuring point and dividing by the standard deviation; constructing a covariance matrix for the normalized data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; selecting the eigenvectors whose cumulative contribution rate reaches a preset threshold as the principal components, and projecting the normalized data into the principal component space to obtain the principal component scores of each measuring point.

[0012] In one possible implementation, hierarchical clustering is used to divide the measuring points into three areas: behind the stop line, above the withdrawal channel, and above the coal wall of the working face, including: The Euclidean distance between any two measuring points based on the principal component scores of pressure, anchor load, top and bottom plate movement and two side movement is calculated; a distance matrix is ​​constructed according to the Euclidean distance, and hierarchical clustering is performed using minimum variance; the optimal number of clusters is determined by calculating the ratio of inter-class distance to intra-class distance, and the measuring points are divided into preset areas.

[0013] In a possible implementation method, a prediction equation of the creep constitutive model is established for the area behind the stop-mining line, including: calculating the roof sinking velocity and acceleration of the measuring points in the area behind the stop-mining line; based on the creep relationship between the roof pressure and the top and bottom plate displacements, using historical data for nonlinear fitting to solve the instantaneous elastic modulus and viscosity coefficient; introducing the surrounding rock damage factor to establish the surrounding rock damage equation including plastic strain and critical strain; constructing the roof pressure prediction equation, and substituting the measured data into solving the initial pressure.

[0014] In one possible implementation, a prediction equation of a viscoelastic-plastic constitutive model is established for the area above the withdrawal channel, including: establishing a viscoelastic relationship equation between the axial stress and displacement of the anchor rod, including the elastic modulus and the viscosity coefficient; introducing a plastic damage term including the softening coefficient and the yield stress, and establishing an anchor rod load prediction equation, including the initial anchor rod load; and using historical data for nonlinear least squares fitting to solve the elastic modulus, viscosity coefficient, softening coefficient and yield stress.

[0015] In one possible implementation, a prediction equation of a rheological damage coupling model is established for the area above the coal wall, including: establishing a rheological equation of stress concentration coefficient and time based on the amount of movement of the two sides, including elastic modulus and viscosity coefficient; introducing a damage evolution equation including damage variables and material parameters, and establishing a stress concentration coefficient prediction equation including an initial stress concentration coefficient; and using historical data for nonlinear fitting to solve all parameters.

[0016] In a surrounding rock damage early warning method based on multi-source data as disclosed above, the embodiment of the present application establishes respective nonlinear prediction models for different data clustering areas, and introduces damage mechanics theory to describe the deterioration process of the surrounding rock. A creep model is introduced in the area behind the stop-mining line to characterize the rheological characteristics of the cantilever beam, a viscoelastic-plastic model is used in the area above the withdrawal channel to describe the softening process of the direct roof, and a rheological damage coupling model is used in the area above the coal wall to characterize the damage evolution of the stress concentration zone, which can more accurately predict the deformation trend of the surrounding rock in each area. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A schematic diagram of a flow chart of a surrounding rock damage early warning method based on multi-source data disclosed in an embodiment of the present application; Figure 2 A schematic diagram of a rear fracture of a stop-mining line disclosed in an embodiment of the present application; Figure 3 A schematic diagram of a rupture above a retracement channel disclosed in an embodiment of the present application; Figure 4 This is a schematic diagram of a fracture above a coal wall disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0019] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that the relative arrangement of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure unless otherwise specifically stated.

[0020] Those skilled in the art can understand that the terms "first", "second" and the like in the embodiments of the present disclosure are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor represent the necessary logical order between them. It should also be understood that in the embodiments of the present disclosure, "multiple" can refer to two or more, and "at least one" can refer to one, two or more. It should also be understood that for any component, data or structure mentioned in the embodiments of the present disclosure, in the absence of explicit limitation or contrary revelation given in the context, it can generally be understood as one or more. In addition, the term "and / or" in the present disclosure is only a description of the association relationship of the associated objects, indicating that there can be three relationships, for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present disclosure generally indicates that the associated objects before and after are an "or" relationship. It should also be understood that the description of each embodiment in the present disclosure emphasizes the differences between the embodiments, and the same or similar parts can refer to each other. For the sake of brevity, they will not be repeated one by one.

[0021] At the same time, it should be understood that, for ease of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship. The following description of at least one exemplary embodiment is actually only illustrative and is by no means intended to limit the present disclosure and its application or use. The techniques, methods and devices known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the techniques, methods and devices should be considered part of the specification. It should be noted that similar numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0022] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] Figure 1 A flowchart of a surrounding rock damage early warning method based on multi-source data disclosed in an embodiment of the present application.

[0024] It should be understood that in the existing technology of surrounding rock damage warning for coal mining working faces, a single data source or a simple linear superposition model is generally used to predict the surrounding rock state. Some methods only establish warning indicators based on displacement monitoring data, without considering other important parameters such as pressure and stress, resulting in one-sided warning results; although some methods collect multi-source data, they use a simple weighted summation method to establish warning models, ignoring the nonlinear characteristics and time effects of surrounding rock deformation. Especially in the surrounding rock warning of the withdrawal channel, due to the advancement of the mining working face, the overburden will produce complex movement deformation laws due to the influence of mining. According to the fracture characteristics of the overburden in different regions, the surrounding rock deformation can be divided into three modes: cantilever beam deformation in the area behind the stop mining line, direct top force deformation above the withdrawal channel, and stress concentration deformation above the coal wall. These three deformation modes have their own unique mechanical characteristics and evolution laws, and it is difficult to accurately describe their deformation mechanisms using traditional single models or linear superposition models. At the same time, during the mining process, the surrounding rock will experience multiple stages such as elastic deformation, plastic softening and damage degradation, and the mechanical response of the rock mass shows significant nonlinearity, time-varying and damage evolution characteristics. Relying only on simple elastic models or linear relationships to predict the state of the surrounding rock cannot accurately reflect these complex mechanical behaviors. Therefore, it is necessary to establish unique nonlinear prediction models for different regions and introduce damage mechanics theory to describe the degradation process of the surrounding rock.

[0025] The embodiment of the present application introduces the Maxwell creep model in the area behind the stop-mining line to characterize the rheological characteristics of the cantilever beam, uses the Kelvin viscoelastic-plastic model in the area above the withdrawal channel to describe the softening process of the immediate roof, and uses the Burgers rheological damage coupling model in the area above the coal wall to characterize the damage evolution of the stress concentration zone, which can more accurately predict the deformation trend of the surrounding rock in each area. This zoning modeling method based on regional characteristics effectively solves the technical problems of single model and linear superposition in the prior art.

[0026] like Figure 1 As shown, in step S101, multiple measuring stations are arranged at predetermined intervals along the retreat channel of the working face, and sensor groups are installed on the roof, solid coal wall, and roadside filling body of the measuring stations. This includes: arranging multiple measuring stations in sequence at preset intervals along the retreat channel of the working face, and installing pressure sensors, axial stress sensors, displacement sensors, and speed sensors at each measuring station; installing pressure sensors on the roof, solid coal wall, and roadside filling body of the working face, and installing axial stress sensors at the end measuring station; collecting and storing the measurement data of the sensors through a data collector to form a data file containing pressure, stress, displacement, and deformation speed.

[0027] In one embodiment, the first measuring station is first arranged at a distance of 20m from the transport chute on the working face, and a measuring station is arranged every 20m along the withdrawal channel. Pressure sensors are used to collect pressure data at different positions of the roof, solid coal wall and roadside filling body, and the sampling frequency is set to once every 2 hours. The first measuring station is arranged at the end, and an axial stress sensor is installed every 20m to collect anchor force data, and the sampling frequency is set to once every 4 hours. Displacement sensors are used to collect displacement data of the roadway roof subsidence and the two sides moving closer, and the sampling frequency is set to once every 6 hours. Velocity sensors are used to collect surrounding rock deformation velocity data, and the sampling frequency is set to once every hour. The measurement range of the sensor is calibrated as follows: the measurement range of the pressure sensor is 0-15MPa; the measurement range of the axial stress sensor is 0-200kN; the measurement range of the displacement sensor is 0-500mm; the measurement range of the velocity sensor is 0-100mm / d. The data measured by all sensors are collected and stored by a data collector to form a data file containing raw data such as pressure, stress, displacement and deformation velocity.

[0028] In step S102, the collected data is processed by using the sliding average method, fast Fourier transform, difference and cumulative summation, and the processed data is aligned according to the measuring point position and time label. This includes: the collected pressure data is smoothed by using the sliding average method, and the spatial distribution characteristics of the pressure data are fitted by the least square method; the axial stress data of the anchor is calculated by the fast Fourier transform method to calculate the spectral characteristics and extract the main frequency component; the displacement data is calculated by the difference method and the time series characteristics are counted; the deformation velocity data is calculated by the cumulative summation method. The processed data is aligned according to the measuring point position and time label to form a processed data table.

[0029] In one embodiment, the pressure data collected at the measuring point is smoothed by the sliding average method, and the window length is set to 24 hours. The least squares method is used to fit the changing trend of the pressure data with the spatial position to obtain the spatial distribution characteristics of the pressure data. For the axial stress data of the anchor rod, the fast Fourier transform method is used to calculate the spectral characteristics of the stress data and extract the main frequency component. For the displacement data, the differential method is used to calculate the displacement increment and the time series characteristics of the displacement data are statistically analyzed. For the deformation velocity data, the cumulative summation method is used to calculate the cumulative deformation. The processed pressure, stress, displacement and deformation velocity data are aligned according to the measuring point position and time label to form a processed data table.

[0030] Taking a certain measuring station of a coal mine withdrawal channel along the gob retention working face as an example, the pressure data of the measuring station changed during multiple days as follows: 2.3MPa, 4.2MPa, 6.6MPa, 8.2MPa, 10.1MPa, 10.1-10.0MPa, 9.8MPa. Fast Fourier transform was performed on the axial stress data of the anchor rod to obtain the main frequency component. The displacement increment was calculated by the differential method for the displacement data. For example, the surrounding rock deformation at 50m ahead of the working face was small, and the top and bottom plates on the real coal body side and the filling body side at 35m approached at 9.2mm / d and 9.5mm / d, respectively, and the approach speed of the two sides was 5.2mm / d. The cumulative summation method was used to calculate the total deformation of the deformation velocity data. All processed data were aligned by measuring points and time to generate a data table.

[0031] In step S103, the aligned data is subjected to principal component analysis and hierarchical clustering, the measuring points are divided into three regions, corresponding prediction models are established respectively, and the model calculation results are normalized. This includes: using principal component analysis to extract the main change characteristics of the data of each measuring point, and calculating the correlation coefficient between each measuring point; using hierarchical clustering to divide the measuring points into three regions: behind the stop mining line, above the withdrawal channel, and above the coal wall of the working face; establishing the prediction equation of the creep constitutive model for the area behind the stop mining line; establishing the prediction equation of the viscoelastic plastic constitutive model for the area above the withdrawal channel; establishing the prediction equation of the rheological damage coupling model for the area above the coal wall; normalizing the predicted values ​​of the three regions to obtain normalized predicted values.

[0032] Principal component analysis is used to extract the main change characteristics of the data of each measuring point, and calculate the correlation coefficient between each measuring point, including: subtracting the measurement mean of the preset time period before the measuring point from the top plate pressure, anchor load, top and bottom plate movement and two-side movement data of each measuring point and dividing by the standard deviation; constructing a covariance matrix for the normalized data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; selecting the eigenvectors whose cumulative contribution rate reaches a preset threshold as the principal components, and projecting the normalized data into the principal component space to obtain the principal component scores of each measuring point.

[0033] Hierarchical clustering is used to divide the measuring points into three areas: behind the stop line, above the withdrawal channel, and above the coal wall of the working face, including: calculating the Euclidean distance between any two measuring points based on the principal component scores of pressure, anchor load, top and bottom plate movement, and two-wall movement; constructing a distance matrix based on the Euclidean distance, and using minimum variance for hierarchical clustering; determining the optimal number of clusters by calculating the ratio of inter-class distance to intra-class distance, and dividing the measuring points into preset areas.

[0034] A prediction equation of the creep constitutive model is established for the area behind the stop-mining line, including: calculating the roof sinking velocity and acceleration of the measuring points in the area behind the stop-mining line; based on the creep relationship between the roof pressure and the top and bottom plate displacements, using historical data for nonlinear fitting to solve the instantaneous elastic modulus and viscosity coefficient; introducing the surrounding rock damage factor to establish the surrounding rock damage equation including plastic strain and critical strain; constructing the roof pressure prediction equation, and substituting the measured data into the solution for the initial pressure.

[0035] A prediction equation of the viscoelastic-plastic constitutive model is established for the area above the withdrawal channel, including: establishing a viscoelastic relationship equation between the axial stress and displacement of the anchor, including the elastic modulus and viscosity coefficient; introducing a plastic damage term including the softening coefficient and yield stress, and establishing an anchor load prediction equation including the initial anchor load; and using historical data for nonlinear least squares fitting to solve the elastic modulus, viscosity coefficient, softening coefficient and yield stress.

[0036] A prediction equation for the rheological damage coupling model is established for the area above the coal wall, including: establishing a rheological equation of stress concentration coefficient and time based on the displacement of the two walls, including elastic modulus and viscosity coefficient; introducing a damage evolution equation including damage variables and material parameters, and establishing a stress concentration coefficient prediction equation including the initial stress concentration coefficient; using historical data for nonlinear fitting to solve all parameters.

[0037] It should be noted that, for identifying the damage characteristics of surrounding rock at different locations, the methods that are widely used at present mainly include deep learning methods based on neural networks, classification methods based on support vector machines, and methods based on gray system theory. Among them, the deep learning method based on neural networks directly trains and predicts the monitoring data by constructing a multi-layer perceptron network structure, but this method requires a large amount of historical data to train the model parameters, and it is often difficult to obtain sufficient training samples in newly excavated mining faces. The classification method based on support vector machines uses kernel functions to map monitoring data to high-dimensional feature space for classification and early warning, but its essence is a statistical classification algorithm that cannot describe the physical mechanism of surrounding rock deformation. The method based on gray system theory predicts the deformation trend of surrounding rock by constructing a GM (1,1) model, but this method assumes that the system has a quasi-exponential development trend, which is inconsistent with the segmented deformation characteristics of surrounding rock in the actual mining process.

[0038] The common technical problem of these methods is that they completely ignore the movement and fracture laws of the overlying rock strata on the mining face, and fail to establish corresponding mechanical models according to the deformation mechanism of the surrounding rock in different regions. The deformation of the surrounding rock under mining conditions is affected by multiple factors such as mining depth, mining thickness, and rock structure, showing obvious spatiotemporal evolution and regional differences. For example, in the area behind the stop-mining line, since the old top rock layer exists in the form of a cantilever beam, the deformation of the surrounding rock is mainly controlled by the rheological properties of the rock layer; in the area above the withdrawal channel, the direct top force causes obvious elastic-plastic deformation of the surrounding rock; in the area above the coal wall, stress concentration causes progressive damage to the surrounding rock. These complex mechanical behaviors cannot be accurately described by a single data-driven model or a simple mathematical model. Therefore, it is necessary to establish prediction equations based on the Maxwell creep model, the Kelvin viscoelastic-plastic model, and the Burgers rheological damage coupling model for the deformation mechanism of the surrounding rock in different regions, and characterize the deformation evolution process of the surrounding rock from the level of physical mechanism. This zoning modeling method based on regional characteristics not only takes into account the mechanical mechanism of surrounding rock deformation under mining conditions, but also calibrates the model parameters through measured data, overcoming the shortcomings of existing early warning methods in both theoretical basis and engineering practice.

[0039] In one embodiment, the principal component analysis method is used to extract the main change characteristics of the data at each measuring point from the data in the processed data table. The correlation coefficient between the pressure and stress, and the correlation coefficient between the displacement and the deformation velocity of each measuring point are calculated. The hierarchical clustering method is used to classify the measuring point data, and the measuring points are divided into three areas: behind the stop line, above the withdrawal channel, and above the coal wall of the working face. Mathematical models are established for the measuring point data in each area: for the measuring points in the area behind the stop line, a cantilever beam force model is established; for the measuring points in the area above the withdrawal channel, a direct top force model is established; for the measuring points in the area above the coal wall, a rock formation fracture model is established. The model calculation results of the three areas are normalized.

[0040] Furthermore, in another embodiment, the measured data is first normalized. For the roof pressure, anchor load, top and bottom plate displacement, and two side displacement data obtained at each measuring point, the measurement mean of the measuring point 168 hours before is subtracted and divided by the standard deviation. A covariance matrix is ​​constructed for the normalized data, and the eigenvalues ​​and eigenvectors are calculated. The eigenvectors with a cumulative contribution rate of 92.7% are selected as the principal components, and the normalized data are projected into the principal component space to obtain the principal component score of each measuring point.

[0041] Perform cluster analysis based on the spatial correlation between measurement points. Calculate the Euclidean distance between any two measurement points i and j: , where P, L, D, and M represent the principal component scores of pressure, anchor load, top and bottom plate displacement, and two-side displacement, respectively. , , , is the corresponding standard deviation. After constructing the distance matrix, the Ward minimum variance method was used for hierarchical clustering. The optimal number of clusters was determined to be 3 by calculating the ratio of the inter-class distance to the intra-class distance. The measuring points were divided into three areas: behind the stop line, above the withdrawal channel, and above the coal wall of the working face.

[0042] For the measuring points in the rear area of ​​the stop mining line, a pressure has occurred before the working face advances to the predetermined position of the withdrawal channel and stops mining. When the withdrawal channel is connected, the next period of pressure in the mining field has not occurred. At this time, the old top rock beam overlying the withdrawal channel is in a stable stage. The structure of the overlying rock layer is as follows: Figure 2 As shown. After the working face stops mining and the overlying roof of the withdrawal channel stabilizes, the rock block A between the coal pillar of the withdrawal channel and the last old roof fracture position can be regarded as the overlying rock block that has an important influence on the activity of the withdrawal channel roof. At this moment, the overlying rock structure of the withdrawal channel can be approximately simplified to a L-long cantilever beam with one end fixed and the other end suspended. A prediction equation based on the Maxwell creep constitutive model can be established. First, calculate the roof sinking velocity vi and acceleration ai at time i in this area. Based on the roof pressure Displacement with top and bottom plates Maxwell creep relation: ,in is the instantaneous elastic modulus, is the viscosity coefficient, and t is the loading time. The measured data are solved by nonlinear fitting and Then the surrounding rock damage factor is introduced ,in is the plastic strain, is the critical strain, and is the material constant. Finally, the prediction equation of the roof pressure in this area is obtained: , represents the sinking speed of the top plate, that is, vi, Represents the sinking acceleration of the top plate, which is ai, and then substitute the historical data for the solution.

[0043] For the measuring points in the area above the withdrawal channel, when the exposed length of the old roof of the working face reaches the pressure step distance, another pressure will occur in the final mining stage. When the pressure step distance is small, the old roof will break above the withdrawal channel, and the spatial structure of the overburden will be as follows: Figure 3 As shown. In this structural state, the rock block B has a significant impact on the withdrawal channel and the support. And the length of the old top periodic pressure step of the overlying broken rock block B is the length L. A prediction equation based on the Kelvin viscoelastic-plastic constitutive model can be established. Anchor bolt axial stress The Kelvin model expression with displacement D is: ,in is the elastic modulus, is the viscosity coefficient. Introducing the plastic damage term , where k is the softening coefficient, is the yield stress. The anchor load prediction equation is established: ,in is the initial anchor load. The measured data are fitted with nonlinear least squares to obtain the parameters , , k and .

[0044] For the measuring points in the area above the coal wall, when the pressure step distance is large, the overlying old roof breaks above the coal wall in front of the working face, and the spatial structure of the overlying rock layer is as follows: Figure 4 As shown. The old roof breaks on the coal body in front of the withdrawal channel. In this structural state, the rock block C with a length of L has a significant impact on the withdrawal channel and the support of the withdrawal channel. A prediction equation based on the Burgers rheological damage coupling model can be established. First, the Burgers creep equation of stress concentration coefficient K and time t is established based on the two-way displacement M: ; in , is the elastic modulus, , is the viscosity coefficient. Introduce the damage evolution equation: ,in is the damage variable, A, n, and m are material parameters. Finally, the stress concentration factor prediction equation is established: . Perform nonlinear fitting on the measured data to solve all parameters.

[0045] The predicted values ​​of the three areas are normalized. For the area behind the stop mining line, the predicted value of roof pressure P is normalized to ; For the area above the retracement channel, the anchor load prediction value L is normalized to ; For the area above the coal wall, the predicted value of the stress concentration factor K is normalized to The subscripts min and max represent the minimum and maximum predicted values ​​of all the measuring points in the region, respectively. Finally, the normalized predicted values ​​of the measuring points in the three regions are obtained, which are used to establish the subsequent early warning indicators.

[0046] At step S104, the abnormal judgment parameters are set according to the normalization processing results, the warning index is calculated by weighted summation, and the warning signal is issued according to the comparison result between the warning index and the preset level threshold. This includes: setting the judgment parameters of pressure abnormality, stress abnormality, displacement abnormality and velocity abnormality; using the weighted summation method to calculate the warning index of each measuring point, and the weight coefficient is determined by experimental calibration; the warning index is divided into preset levels according to the numerical value, and a warning signal is issued when the warning index exceeds the corresponding level threshold.

[0047] Specifically, according to the normalized model calculation results, the warning parameters are set: when the pressure value of the measuring point is greater than 1.5 times the average pressure value of the measuring point in the previous 24 hours, it is recorded as a pressure anomaly; when the stress value of the measuring point is greater than 1.8 times the average stress value of the measuring point in the previous 24 hours, it is recorded as a stress anomaly; when the displacement increment of the measuring point is greater than 2.0 times the average displacement increment of the measuring point in the previous 24 hours, it is recorded as a displacement anomaly; when the deformation speed of the measuring point is greater than 2.2 times the average deformation speed of the measuring point in the previous 24 hours, it is recorded as a speed anomaly. The weighted summation method is used to calculate the warning index of each measuring point, and the weight coefficient is determined by experimental calibration. The warning index is divided into three levels according to the numerical value, and a warning signal is issued when the warning index exceeds the corresponding level threshold.

[0048] In one embodiment, a coal mine working face as described above is taken as an example: when the pressure value at a measuring point exceeds 1.5 times the average pressure value of the point in the previous 24 hours, it is recorded as a pressure anomaly. For example, if the average pressure value at a measuring point in the previous 24 hours is 6.0 MPa, it is recorded as an anomaly when the measured pressure value is greater than 9.0 MPa; when the stress value at a measuring point exceeds 1.8 times the average stress value at the point in the previous 24 hours, it is recorded as a stress anomaly. For example, if the average stress value at a measuring point in the previous 24 hours is 50 kN, it is recorded as an anomaly when the measured stress value is greater than 90 kN; when the displacement increment at a measuring point exceeds 2.0 times the average displacement increment value at the point in the previous 24 hours, it is recorded as a displacement anomaly. For example, if the average displacement increment value at a measuring point in the previous 24 hours is 5 mm / d, it is recorded as an anomaly when the measured displacement increment is greater than 10 mm / d; when the deformation speed at a measuring point exceeds 2.2 times the average deformation speed value at the point in the previous 24 hours, it is recorded as a speed anomaly. For example, if the average deformation speed at a measuring point in the previous 24 hours is 8 mm / d, it is recorded as an anomaly when the measured deformation speed is greater than 17.6 mm / d. The weight coefficient of each abnormal item is determined through field test calibration, and the weights of pressure, stress, displacement and velocity are 0.3, 0.3, 0.2 and 0.2 respectively. The warning index is divided into three levels according to the range of 0-1.0, 0-0.6 is the first level warning, 0.6-0.8 is the second level warning, and 0.8-1.0 is the third level warning.

[0049] Furthermore, an embodiment of the present application also discloses a surrounding rock damage warning device, including: a processor, a memory, and a system bus; the processor and the memory are connected via the system bus; the memory is used to store one or more programs, and the one or more programs include instructions, and when the instructions are executed by the processor, the processor executes any of the above methods.

[0050] Furthermore, an embodiment of the present application also discloses a computer program product, which, when executed on a terminal device, enables the terminal device to execute any of the above methods.

[0051] It can be known from the description of the above implementation mode that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for enabling a computer device (which can be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0052] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0053] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0054] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A surrounding rock damage early warning method based on multi-source data, characterized in that: include: A plurality of measuring stations are arranged at predetermined intervals along the retreat channel of the working face, and sensor groups are installed on the roof, solid coal wall, and filling bodies beside the roadway at the measuring stations; The collected data are processed using sliding average method, fast Fourier transform, difference and cumulative summation, and the processed data are aligned according to the measurement point location and time label; Performing principal component analysis and hierarchical clustering on the aligned data, dividing the measurement points into three regions, establishing corresponding prediction models respectively, and normalizing the model calculation results; The abnormal judgment parameters are set according to the normalized processing results, the warning index is calculated by weighted summation, and the warning signal is issued based on the comparison result between the warning index and the preset level threshold.

2. The surrounding rock damage early warning method according to claim 1, characterized in that: in, A plurality of measuring stations are arranged at predetermined intervals along the retreat channel of the working face, and a sensor group is installed on the roof, solid coal wall, and filling body beside the roadway at the measuring station, including: Multiple measuring stations are arranged in sequence along the retreat channel of the working face at preset intervals, and each measuring station is equipped with a pressure sensor, an axial stress sensor, a displacement sensor and a velocity sensor; Pressure sensors are installed on the working face roof, solid coal wall and roadside filling respectively, and axial stress sensors are installed at the end measuring station; The measurement data of the sensor is collected and stored by a data collector to form a data file containing pressure, stress, displacement and deformation speed.

3. The surrounding rock damage early warning method according to claim 1, characterized in that: in, The collected data is processed using sliding average method, fast Fourier transform, difference and cumulative summation, and the processed data is aligned according to the measurement point location and time label, including: The collected pressure data are smoothed by the sliding average method, and the spatial distribution characteristics of the pressure data are fitted by the least square method; The fast Fourier transform method is used to calculate the spectrum characteristics of the anchor axial stress data and extract the main frequency component; The displacement data is subjected to differential method to calculate displacement increment and statistical time series characteristics; The cumulative deformation is calculated by using the cumulative summation method for the deformation velocity data; The processed data are aligned according to the measurement point locations and time tags to form a processed data table.

4. The surrounding rock damage early warning method according to claim 1, characterized in that: in, The aligned data are subjected to principal component analysis and hierarchical clustering, the measurement points are divided into three regions, corresponding prediction models are established respectively, and the model calculation results are normalized, including: The principal component analysis is used to extract the main variation characteristics of the data at each measuring point and calculate the correlation coefficient between the measuring points; Hierarchical clustering was used to divide the measuring points into three areas: behind the stop-mining line, above the withdrawal channel, and above the coal wall of the working face. The prediction equation of creep constitutive model is established for the area behind the stop-mining line; The prediction equation of the viscoelastic-plastic constitutive model is established for the area above the retracement channel; The prediction equation of rheological damage coupling model is established for the area above the coal wall; The predicted values ​​of the three regions are normalized to obtain normalized predicted values.

5. The surrounding rock damage early warning method according to claim 1, characterized in that: in, According to the normalized processing results, the abnormal judgment parameters are set, and the warning index is calculated by weighted summation. The warning signal is issued according to the comparison result between the warning index and the preset level threshold, including: Set the judgment parameters for pressure anomaly, stress anomaly, displacement anomaly and velocity anomaly; The weighted summation method is used to calculate the early warning index of each measuring point, and the weight coefficient is determined through experimental calibration; The warning index is divided into preset levels according to its numerical value, and a warning signal is issued when the warning index exceeds the corresponding level threshold.

6. The surrounding rock damage early warning method according to claim 4, characterized in that: in, The principal component analysis is used to extract the main change characteristics of the data at each measuring point and calculate the correlation coefficients between the measuring points, including: The roof pressure, anchor load, top and bottom plate displacement and two side displacement data of each measuring point are respectively deducted from the measurement mean value of the preset time period before the measuring point and then divided by the standard deviation; Constructing a covariance matrix for the normalized data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; The eigenvectors whose cumulative contribution rate reaches the preset threshold are selected as principal components, and the normalized data are projected into the principal component space to obtain the principal component scores of each measuring point.

7. The surrounding rock damage early warning method according to claim 4, characterized in that: in, Hierarchical clustering is used to divide the measuring points into three areas: behind the stop-mining line, above the withdrawal channel, and above the coal wall of the working face, including: Calculate the Euclidean distance between any two measuring points based on the principal component scores of pressure, anchor load, top and bottom plate displacement, and two side displacement; Constructing a distance matrix according to the Euclidean distance, and performing hierarchical clustering using minimum variance; The optimal number of clusters is determined by calculating the ratio of the inter-class distance to the intra-class distance, and the measurement points are divided into preset areas.

8. The surrounding rock damage early warning method according to claim 4, characterized in that: in, The prediction equation of the creep constitutive model is established for the area behind the stop-mining line, including: Calculate the roof sinking speed and acceleration of the measuring points in the area behind the stop mining line; Based on the creep relationship between top plate pressure and top and bottom plate displacement, the instantaneous elastic modulus and viscosity coefficient are solved by nonlinear fitting using historical data; The surrounding rock damage factor is introduced to establish the surrounding rock damage equation including plastic strain and critical strain; The roof pressure prediction equation was constructed and the measured data were substituted into the equation to solve for the initial pressure.

9. The surrounding rock damage early warning method according to claim 4, characterized in that: in, The prediction equation of the viscoelastic-plastic constitutive model is established for the area above the retracement channel, including: Establish the viscoelastic relationship equation between the axial stress and displacement of the anchor rod, including the elastic modulus and viscosity coefficient; The plastic damage term including softening coefficient and yield stress is introduced, and the anchor load prediction equation is established, including the initial anchor load; The elastic modulus, viscosity coefficient, softening coefficient and yield stress are solved by nonlinear least squares fitting using historical data.

10. The surrounding rock damage early warning method according to claim 4, characterized in that: in, The prediction equation of the rheological damage coupling model is established for the area above the coal wall, including: According to the displacement of the two sides, a rheological equation of stress concentration factor and time is established, including elastic modulus and viscosity coefficient; The damage evolution equation is introduced, including damage variables and material parameters, and the stress concentration factor prediction equation is established, including the initial stress concentration factor; All parameters are solved by nonlinear fitting using historical data.

Images