Real-time analysis method for stress distribution of surrounding rock of deep complex roadway

Abstract

The invention discloses a deep complex roadway surrounding rock stress distribution real-time analysis method, which comprises the following steps: dividing a to-be-detected area, and scanning to obtain surrounding rock stress change data information in real time; analyzing and processing the data information, inputting the data information into a model to generate a surrounding rock stress change coefficient, and evaluating the surrounding rock stress change condition; based on a machine learning model evaluation result, surrounding rock stress change division is perceived, and a stress change evaluation result is obtained; for an evaluation result, continuously obtaining surrounding rock stress change information, and dividing the surrounding rock stress change information into different levels of risk stress changes; and selecting different early warning measures according to different levels of risk stress changes. According to the method, the to-be-detected area is divided, data are collected in real time, the surrounding rock stress change coefficient is combined, abnormal stress change is accurately recognized, and the detection capacity is improved. After abnormal changes are detected, multi-window grading evaluation is carried out, and corresponding early warning is sent out, so that an early warning system is optimized, and construction safety and response efficiency are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of surrounding rock stress analysis, and particularly to a real-time analysis method for the stress distribution of surrounding rock in deep and complex roadways. Background Technique

[0002] The real-time analysis of the stress distribution of surrounding rock in deep and complex roadways is to, during the underground roadway mining or construction process, through advanced sensor and data processing technologies, monitor and analyze the stress state of the surrounding rock in real time, helping engineers understand the stress conditions, deformations, failure risks of the rock and their impacts on the roadway stability. By obtaining stress data in real time, engineers can timely identify potential dangerous areas and take preventive and reinforcement measures to ensure construction safety and the long-term stability of the roadway. In addition, this technology provides important decision-making support for the design and construction of deep mines or tunnels. By combining historical data and on-site monitoring results, and using data mining and machine learning algorithms for in-depth analysis, a comprehensive understanding of the mechanical properties of the surrounding rock can be achieved, stress concentration phenomena can be predicted, and the impacts of different mining or construction schemes can be evaluated. This feedback mechanism based on real-time data improves the safety of the project and provides new ideas for optimizing the construction process, reducing costs and minimizing resource waste, becoming an essential and important part of modern underground project management.

[0003] The real-time analysis of the stress distribution of surrounding rock in deep and complex roadways can utilize advanced sensor technologies, including three-dimensional laser scanners. This device can accurately capture the three-dimensional spatial data of the roadway and its surrounding surrounding rock, generating a high-resolution point cloud model, providing a detailed geometric basis for subsequent stress analysis. By combining with other sensors (such as strain gauges and displacement sensors), the three-dimensional laser scanner can monitor the deformation and stress changes of the surrounding rock in real time, identifying stress concentration and potential failure areas. This data integration and analysis helps to establish a mechanical model of the surrounding rock, enabling engineers to more accurately simulate and predict the stress distribution, thereby timely discovering potential safety hazards. In addition, the fast data acquisition ability of laser scanning improves the monitoring efficiency. By combining real-time data processing technologies and machine learning algorithms, dynamic monitoring of the stress state of the surrounding rock is achieved, the construction plan is optimized, and the safety and stability of the roadway are ensured. The integration of this technology marks the intelligent development of modern underground project management and provides strong technical support for the safe mining of deep and complex roadways.

[0004] The existing technologies have the following deficiencies:

[0005] Existing 3D laser scanners usually adopt a fixed field of view angle when collecting surrounding rock data. This design aims to meet the needs of most common usage scenarios to ensure that a predetermined area can be covered and sufficient point cloud data can be obtained. However, there may be complex fracture networks inside the surrounding rock, and the fixed field of view angle may not be able to effectively capture all fractures, especially the tiny fractures located at the edges or deep inside. Such omissions will seriously affect the assessment of the overall stability of the surrounding rock, thus threatening construction safety. At the same time, the presence of fracture networks will change the stress distribution characteristics of the surrounding rock. If these fractures are not detected in time, it may lead to misjudgment of the stress state of the surrounding rock, and then make the design and reinforcement measures ineffective, exacerbating the instability of the surrounding rock. Summary of the Invention

[0006] The object of the present invention is to overcome the deficiencies existing in the prior art. To achieve the above object, a real-time analysis method for the stress distribution of surrounding rock in deep and complex roadways is adopted to solve the problems raised in the above background technology.

[0007] A real-time analysis method for the stress distribution of surrounding rock in deep and complex roadways includes the following steps:

[0008] Step S1: Divide the area to be detected of the surrounding rock of the roadway into multiple sub-areas, and obtain the data information of the stress change of the surrounding rock in each sub-area in real time through scanning. The data information includes the distribution of the normal vector directions of each point on the surface of the surrounding rock and the change in the roughness of the surface of the surrounding rock.

[0009] Step S2: Analyze and process the data information obtained for each sub-area and then input it into a pre-trained machine learning model. Generate a stress change coefficient of the surrounding rock through the machine learning model, and conduct an intelligent evaluation of the stress change situation of the surrounding rock based on the stress change coefficient.

[0010] Step S3: Based on the evaluation results of the machine learning model, divide the stress change of the surrounding rock into normal stress change or abnormal stress change, conduct an intelligent perception of the stress change of the surrounding rock, and obtain a stress change evaluation result.

[0011] Step S4: For the evaluated abnormal stress change, continuously obtain the stress change information of the surrounding rock, and divide the abnormal stress change into different levels of risk stress changes, including low-risk stress change, medium-risk stress change, and high-risk stress change.

[0012] Step S5: Select different warning measures according to the divided different levels of risk stress changes.

[0013] As a further solution of the present invention: The distribution of the normal vector directions of the surface of the surrounding rock is the direction and distribution characteristics of the normal vector of each point on the surface of the surrounding rock, and the change in the roughness of the surface of the surrounding rock is the change in the smooth or rough degree of the surface of the surrounding rock.

[0014] As a further solution of the present invention: The specific steps for analyzing and processing the data information obtained for each sub-region include:

[0015] Analyze the obtained normal direction distribution on the surrounding rock surface and the change in the roughness of the surrounding rock surface, respectively generate a normal direction distribution index and a surface roughness index, and input the generated normal direction distribution index and surface roughness index into a pre-trained machine learning model.

[0016] As a further solution of the present invention: After analyzing the normal direction distribution on the surrounding rock surface, a normal direction distribution index is generated. The specific steps are as follows:

[0017] Scan to obtain the point cloud data on the surrounding rock surface, and calibrate the obtained point cloud data as:

[0018] where P i is the i-th measurement point on the surrounding rock surface, and each point P i contains three coordinate values, denoted as {x i , y i , z i}, where x i , y i , z i respectively represent the x, y, and z-axis coordinates of P i in three-dimensional space, and n is the total number of measurement points, that is, the total number of points obtained during the scanning process;

[0019] By analyzing the neighborhood points around each point, calculate the normal vector of each point. The calculation expression is as follows:

[0020]

[0021] In the formula, N i is the normal vector of the surrounding rock surface point P i , N(i) represents the set of neighborhood points of the surrounding rock surface point P i , that is, a group of adjacent points around the surrounding rock surface point P i , P j is the j-th measurement point on the surrounding rock surface, and w j is the weight coefficient of the surrounding rock surface point P j ;

[0022] Calculate the angles of each normal vector N i with respect to the x, y, and z axes. The calculation expression is as follows:

[0023]

[0024] In the formula, θ i,x , θ i,y and θi,y They are the normal vector N i For the angles with respect to the x, y, and z axes;

[0025] The angle θ in the normal direction i Is classified into predefined angle intervals to generate a normal direction histogram, and the calculation expression is as follows:

[0026]

[0027] In the formula, H x (k), H y (k) and H z (k) are respectively the angle distribution histograms of the normal direction with respect to the x, y, and z axes, and respectively count the number of all normal vectors within the angle interval I k Among them, I k Is the k-th angle interval, representing an angle range used to classify different normal angles into corresponding intervals, And Among them, Is an indicator function used to judge whether the angle θ i,x , θ i,y Or θ i,y Is within the interval I k Inside;

[0028] Calculate the entropy values of the angle distribution histograms H x (k), H y (k) and H z (k) to generate a normal direction distribution index, and the calculation expression is as follows:

[0029]

[0030] In the formula, NDDI is the normal direction distribution index, a ∈ {x, y, z} is the summation for a, a represents the three coordinate axes x, y, z, m is the total number of angle intervals, and H a (k) is the number of normal vectors within the k-th angle interval I k For the coordinate axis a.

[0031] As a further solution of the present invention: After analyzing the roughness change of the surrounding rock surface, a surface roughness index is generated, and the specific steps are as follows:

[0032] Under the detection window, obtain the point cloud data of the surrounding rock surface from a 3D laser scanner, including the three-dimensional coordinates x, y, z of each point P i , and select each measurement point P iFor the neighborhood point set of , calculate the height difference of the measurement point relative to its neighborhood points, and record the maximum height difference and the minimum height difference. The calculation expression is as follows:

[0033]

[0034] In the formula, ΔH i represents the local height difference of the i-th measurement point, and N(i) represents the neighborhood point set of the surrounding rock surface point P i ; is the height value of the highest point in the neighborhood point set N(i), and

[0035] is the height value of the lowest point in the neighborhood point set N(i); According to the local height difference, calculate the roughness change rate of each point and correlate it with the specific position distance of the detection point i. The calculation expression is as follows:

[0036]

[0037] In the formula, R change is the roughness change rate, indicating the roughness change degree at the measurement point i, and D i is the neighborhood average distance of the measurement point i;

[0038] Perform a non-linear mapping on the roughness change rate to generate the roughness index of each point. The calculation expression is as follows:

[0039]

[0040] In the formula, R i is the roughness degree of the point P i , e is the natural base, B is the adjustment parameter used to control the response sensitivity of the roughness index to the change rate, and β is the threshold value used to determine the critical value of the roughness change rate;

[0041] Perform a weighted average on the roughness indices of all points to generate the final surface roughness index. The calculation expression is as follows:

[0042]

[0043] In the formula, SRI is the surface roughness index, n is the total number of measurement points, and F i is the weight of the measurement point i.

[0044] As a further solution of the present invention: The surrounding rock change coefficient RMVC is generated by the machine learning model, and the formula is as follows:

[0045]

[0046] In the formula, k 1 、k2 They are preset proportionality coefficients of the normal direction distribution index NDDI and the surface roughness index SRI respectively, and k 1 and k 2 are both greater than 0.

[0047] As a further solution of the present invention: the division of the surrounding rock stress change includes the following steps:

[0048] If the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the surrounding rock change coefficient, the surrounding rock stress change is divided into abnormal stress change;

[0049] If the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient, the surrounding rock stress change is divided into normal stress change.

[0050] As a further solution of the present invention: for the evaluated abnormal stress change, continuously obtain the surrounding rock stress change information, and the specific steps include:

[0051] By calculating the average value and standard deviation of several surrounding rock change coefficients in the analysis set, and comparing and analyzing the obtained average value of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient with the preset reference threshold of the average value of the surrounding rock change coefficient and the preset reference threshold of the standard deviation of the surrounding rock change coefficient respectively, further divide the abnormal stress change, and the division process is as follows:

[0052] If the average value of the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the surrounding rock change coefficient, the abnormal stress change is further divided into high-risk stress change;

[0053] If the average value of the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the standard deviation of the surrounding rock change coefficient, the abnormal stress change is further divided into medium-risk stress change;

[0054] If the average value of the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient is less than the preset reference threshold of the standard deviation of the surrounding rock change coefficient, the abnormal stress change is further divided into low-risk stress change.

[0055] Compared with the prior art, the present invention has the following technical effects:

[0056] With the above technical solution, by dividing the area to be detected into multiple sub-areas and using a three-dimensional laser scanner to collect the data of the surrounding rock stress changes in each sub-area in real time, the monitoring range is more comprehensive and detailed. In particular, it can capture the tiny cracks in the deep and edge areas. This detailed data collection method effectively avoids the risk of information omission caused by the traditional fixed field of view angle. Using the normal direction distribution index and roughness index generated from the normal direction distribution and surface roughness changes of the surrounding rock surface, through the processing of a machine learning model, a surrounding rock stress change coefficient is generated to conduct an intelligent evaluation of the stress change. This process can accurately judge the abnormal stress changes, improve the recognition ability of complex fracture networks, and thus better evaluate the overall stability of the surrounding rock, providing a strong guarantee for construction safety.

[0057] After the present invention detects abnormal stress changes, the system continuously obtains the surrounding rock stress change information, generates a data set through multi-window detection, calculates the average value and standard deviation of the surrounding rock change coefficient, compares with the reference threshold, and further divides it into low-risk, medium-risk, and high-risk stress changes. This hierarchical evaluation method can accurately distinguish stress changes of different risk levels, issue red and yellow warnings for high-risk and medium-risk changes respectively, and record low-risk changes in detail for subsequent analysis. This hierarchical warning mechanism not only avoids unnecessary alarm interference but also ensures a rapid response to high-risk events, optimizing the effectiveness of the warning system, enabling construction management personnel to take corresponding measures according to the risk level, and ensuring the safety and continuity of the construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings:

[0059] Figure 1 It is a schematic diagram of the steps of the real-time analysis method for the disclosed embodiment of the present application. SPECIFIC EMBODIMENTS

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0061] Please refer to Figure 1 , in the embodiment of the present invention, a real-time analysis method for the stress distribution of the surrounding rock in a deep and complex roadway includes the following steps:

[0062] Step S1: Divide the area to be detected of the roadway surrounding rock into multiple sub - areas, and use a 3D laser scanner to scan and obtain the data information of the stress change of the surrounding rock in each sub - area in real - time. The data information includes the distribution of the normal direction on the surface of the surrounding rock and the change in the roughness of the surface of the surrounding rock;

[0063] Specifically, for a systematic analysis of a relatively large surrounding - rock area, it is divided into several smaller parts for more detailed monitoring. This method helps to improve the accuracy and efficiency of monitoring. Since the characteristics and conditions of each sub - area may vary, by detecting each sub - area one by one, the overall condition of the surrounding rock in the whole area can be understood more comprehensively.

[0064] Obtaining the data information of the stress change of the surrounding rock in each sub - area in real - time by a 3D laser scanner means using laser scanning technology to collect and analyze the stress data of the surrounding rock in each divided sub - area in real - time. The 3D laser scanner can quickly obtain high - precision point - cloud data, providing detailed information about the surface shape, structure, and stress state of the surrounding rock. This real - time data - acquisition method enables engineers to monitor the dynamic changes of the surrounding rock immediately, identify potential safety hazards in a timely manner, and thus effectively guide subsequent construction and maintenance work.

[0065] In this embodiment, the distribution of the normal direction on the surface of the surrounding rock is the direction and distribution characteristics of the normal vector of each point on the surface of the surrounding rock, and the change in the roughness of the surface of the surrounding rock is the change in the smooth or rough degree of the surface of the surrounding rock.

[0066] Step S2: After analyzing and processing the data information obtained for each sub - area, input it into a pre - learned machine - learning model. Generate the surrounding - rock stress change coefficient through the machine - learning model, and conduct an intelligent evaluation of the surrounding - rock stress change situation based on the surrounding - rock change coefficient;

[0067] In this embodiment, the specific steps for analyzing and processing the data information obtained for each sub - area include:

[0068] Analyze the obtained distribution of the normal direction on the surface of the surrounding rock and the change in the roughness of the surface of the surrounding rock, generate the normal - direction distribution index and the surface - roughness index respectively, and input the generated normal - direction distribution index and surface - roughness index into a pre - learned machine - learning model.

[0069] The uneven distribution of the normal direction on the surrounding rock surface may indeed indicate the existence of a complex fracture network within the surrounding rock. This is because the change in the normal direction reflects the subtle differences in the shape and structure of the surrounding rock surface. When there are fractures within the surrounding rock, these fractures will cause local deformation of the rock mass under stress, thereby affecting the normal direction of the surface. If the fracture distribution is complex and irregular, the normal direction of the surrounding rock surface will exhibit obvious scatter and unevenness. Especially in the fracture concentration area or the intersection area, the direction of the normal may change drastically. In addition, the uneven distribution of the normal direction may also affect the stress transfer and deformation characteristics of the surrounding rock, further leading to stress concentration and potential instability in local areas. Therefore, by monitoring the distribution of the normal direction, the complexity of the fracture network within the surrounding rock can be indirectly inferred, providing an important basis for subsequent safety assessment and reinforcement measures.

[0070] In this embodiment, after analyzing the distribution of the normal direction on the surrounding rock surface, a normal direction distribution index is generated. The specific steps are as follows:

[0071] Under the detection window, the point cloud data of the surrounding rock surface is obtained by scanning with a three-dimensional laser scanner, and the obtained point cloud data is calibrated as:

[0072] where P i is the i-th measurement point on the surrounding rock surface, and each point P i contains three coordinate values, denoted as {x i , y i , z i}, where x i , y i , z i respectively represent the x, y, and z-axis coordinates of P i in three-dimensional space, and n is the total number of measurement points, that is, the total number of points obtained during the scanning process;

[0073] By analyzing the neighboring points around each point, the normal vector of each point is calculated. The calculation expression is as follows:

[0074]

[0075] In the formula, N i is the normal vector of the surrounding rock surface point P i , N(i) represents the set of neighboring points of the surrounding rock surface point P i , that is, a group of adjacent points around the surrounding rock surface point P i , P j is the j-th measurement point on the surrounding rock surface, and w j is the weight coefficient of the surrounding rock surface point P j ;

[0076] Calculate each normal vector Ni For the angles of the x, y, and z axes, the calculation expressions are as follows:

[0077]

[0078] In the formula, θ i,x 、θ i,y and θ i,y are the angles of the normal vector N i with respect to the x, y, and z axes respectively;

[0079] Classify the angle θ i in the normal direction into predefined angle intervals to generate a normal direction histogram. The calculation expression is as follows:

[0080]

[0081] In the formula, H x (k), H y (k) and H z (k) are the angle distribution histograms of the normal direction with respect to the x, y, and z axes respectively, and count the number of all normal vectors within the angle interval I k for the angles between the normal vectors and the x, y, and z axes. I k is the k-th angle interval, representing an angle range used to classify different normal angles into corresponding intervals, and in, is an indicator function used to determine whether the angle θ i,x , θ i,y or θ i,y is within the interval I k ;

[0082] Calculate the entropy value of the angle distribution histograms H x (k), H y (k) and H z (k) of the x, y, and z axes to generate a normal direction distribution index. The calculation expression is as follows:

[0083]

[0084] In the formula, NDDI is the normal direction distribution index, a ∈ {x, y, z} is the sum for a, a represents the three coordinate axes x, y, z, m is the total number of angle intervals, and H a (k) is the number of normal vectors in the k-th angle interval I k for the coordinate axis a.

[0085] From the calculation expression of the normal direction distribution index, it can be seen that under the detection window, the larger the value of the normal direction distribution index generated after analyzing the normal direction distribution on the surrounding rock surface, the more obvious the change in the normal direction of the surrounding rock surface, indicating that there may be more cracks or structural defects inside the surrounding rock. This non-uniformity reflects a potential complex crack network. On the contrary, if the value of the normal direction distribution index is small, it means that the change in the normal direction is relatively uniform, indicating that the surface of the surrounding rock is relatively flat and stable, thus reducing the probability of a complex crack network inside.

[0086] The abnormal change in the roughness of the surrounding rock surface can indeed indicate that there may be a complex crack network inside the surrounding rock. This is because the surface state of the surrounding rock directly reflects its internal structure and characteristics. When there is a complex crack network inside the surrounding rock, these cracks will cause local deformation and displacement of the surrounding rock when it is stressed or the environment changes, thus affecting its surface roughness. Specifically, the existence of cracks may cause the surface of some areas to be uneven, resulting in obvious roughness changes. In addition, the existence of the crack network will also affect the mechanical properties of the surrounding rock, causing stress concentration and uneven distribution, thus further exacerbating the change in surface roughness. This abnormal fluctuation of roughness is often an indication of the complexity and instability of the internal structure of the surrounding rock, predicting potential safety hazards. Therefore, monitoring and analyzing the change in the roughness of the surrounding rock surface is crucial for evaluating the stability of the surrounding rock, and can provide important clues for engineers to take timely measures to prevent possible disasters such as collapses or landslides.

[0087] In this embodiment, after analyzing the change in the roughness of the surrounding rock surface, a surface roughness index is generated. The specific steps are as follows:

[0088] Under the detection window, obtain the point cloud data of the surrounding rock surface from the 3D laser scanner, including the three-dimensional coordinates x, y, z of each point P i , select the neighborhood point set of each measurement point P i , calculate the height difference between the measurement point and its neighborhood points, and record the maximum height difference and the minimum height difference. The calculation expression is as follows:

[0089]

[0090] In the formula, ΔH i represents the local height difference of the i-th measurement point, N(j) represents the neighborhood point set of the surrounding rock surface point P i , is the height value of the highest point in the neighborhood point set N(i), is the height value of the lowest point in the neighborhood set N(i);

[0091] According to the local height difference, the roughness change rate of each point is associated with the distance from the specific position of the detection point i, and the calculation expression is as follows:

[0092]

[0093] In the formula, R change is the roughness change rate, indicating the degree of roughness change at the measurement point i, and D i is the average distance of the neighborhood of the measurement point i;

[0094] The roughness change rate is non-linearly mapped to generate the roughness index of each point, and the calculation expression is as follows:

[0095]

[0096] In the formula, R i is the roughness of point P i , e is the natural base, B is the adjustment parameter, used to control the response sensitivity of the roughness index to the change rate, and β is the threshold, used to determine the critical value of the roughness change rate;

[0097] The roughness indices of all points are weighted and averaged to generate the final surface roughness index, and the calculation expression is as follows:

[0098]

[0099] In the formula, SRI is the surface roughness index, n is the total number of measurement points, and F i is the weight of the measurement point i.

[0100] It can be seen from the calculation expression of the surface roughness index that under the detection window, the larger the value of the surface roughness index generated after analyzing the roughness change of the surrounding rock surface usually means that there are more unevenness and irregular features on the surrounding rock surface, which is often a reflection of the internal fracture network. These fractures may cause the surrounding rock to deform when stressed, and thus produce significant roughness changes on the surface. Therefore, a higher roughness index usually indicates a greater probability of the existence of a complex fracture network. On the contrary, if the surface roughness index is small, it indicates that the surrounding rock surface is relatively smooth and regular, which may mean that there are fewer or simpler internal fractures, thus reducing the possibility of the existence of a complex fracture network.

[0101] In this embodiment, the surrounding rock change coefficient RMVC is generated by the machine learning model, and the formula is as follows:

[0102]

[0103] In the formula, k 1 , k 2They are preset proportionality coefficients of the normal direction distribution index NDDI and the surface roughness index SRI respectively, and k 1 and k 2 are both greater than 0.

[0104] It can be seen from the calculation expression of the surrounding rock change coefficient that under the detection window, the larger the value of the normal direction distribution index generated after analyzing the normal direction distribution of the surrounding rock surface, and the larger the value of the surface roughness index generated after analyzing the roughness change of the surrounding rock surface, it indicates that the larger the value of the surrounding rock change coefficient generated under the monitoring window, and the greater the probability that there may be a complex fracture network inside the surrounding rock. On the contrary, it indicates that the probability that there may be a complex fracture network inside the surrounding rock is smaller.

[0105] Step S3: Based on the evaluation result of the machine learning model, divide the surrounding rock stress change into normal stress change or abnormal stress change, and intelligently perceive the surrounding rock stress change to obtain the stress change evaluation result;

[0106] In this embodiment, the division of the surrounding rock stress change includes the following steps:

[0107] If the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the surrounding rock change coefficient, then divide the surrounding rock stress change into abnormal stress change;

[0108] If the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient, then divide the surrounding rock stress change into normal stress change.

[0109] Step S4: For the evaluated abnormal stress change, continuously obtain the surrounding rock stress change information, and divide the abnormal stress change into different levels of risk stress changes, including low-risk stress change, medium-risk stress change, and high-risk stress change;

[0110] In this embodiment, for the evaluated abnormal stress change, continuously obtain the surrounding rock stress change information, and conduct a comprehensive analysis by establishing a data set of the surrounding rock change coefficients generated under several detection windows. The specific steps include:

[0111] By calculating the average value and standard deviation of several surrounding rock change coefficients in the analysis set, and comparing and analyzing the obtained average value of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient with the preset reference threshold of the average value of the surrounding rock change coefficient and the preset reference threshold of the standard deviation of the surrounding rock change coefficient respectively, further divide the abnormal stress change. The division process is as follows:

[0112] If the average value of the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the surrounding rock change coefficient, then further divide the abnormal stress change into high-risk stress change;

[0113] If the average value of the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient is greater than or equal to the preset reference threshold of the standard deviation of the surrounding rock change coefficient, the abnormal stress change is further divided into medium-risk stress change;

[0114] If the average value of the surrounding rock change coefficient is less than the preset reference threshold of the surrounding rock change coefficient and the standard deviation of the surrounding rock change coefficient is less than the preset reference threshold of the standard deviation of the surrounding rock change coefficient, the abnormal stress change is further divided into low-risk stress change.

[0115] Step S5: Select different warning measures according to the classified risk stress changes of different levels.

[0116] In the present invention, the area to be detected is divided into multiple sub-areas, and the surrounding rock stress change data of each sub-area are collected in real time by using a three-dimensional laser scanner, so that the monitoring range is more comprehensive and detailed. In particular, it can capture the tiny cracks in the depth and at the edges. This detailed data collection method effectively avoids the risk of information omission caused by the traditional fixed field of view angle. By using the normal direction distribution index and roughness index generated from the normal direction distribution on the surface of the surrounding rock and the change of surface roughness, and through the processing of a machine learning model, the surrounding rock stress change coefficient is generated to conduct an intelligent evaluation of the stress change. This process can accurately judge the abnormal stress change, improve the recognition ability of complex fracture networks, and thus better evaluate the overall stability of the surrounding rock, providing a strong guarantee for construction safety.

[0117] After the present invention detects an abnormal stress change, the system continuously obtains the surrounding rock stress change information, generates a data set through multi-window detection, calculates the average value and standard deviation of the surrounding rock change coefficient, compares them with the reference threshold, and further divides them into low-risk, medium-risk and high-risk stress changes. This hierarchical evaluation method can accurately distinguish the stress changes of different risk levels, issue red and yellow warnings for high-risk and medium-risk changes respectively, and record the low-risk changes in detail for subsequent analysis. This hierarchical warning mechanism not only avoids unnecessary alarm interference, but also ensures a rapid response to high-risk events, optimizes the effectiveness of the warning system, enabling construction management personnel to take corresponding measures according to the risk level to ensure the safety and continuity of the construction process.

[0118] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents, and all should be included within the protection scope of the present invention.

Claims

1. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels, characterized in that: The following steps are involved: Step S1, dividing the area to be detected of the tunnel surrounding rock into multiple sub-areas, and obtaining the surrounding rock stress change data information of each sub-area in real time by scanning, wherein the data information includes the normal direction distribution of the surrounding rock surface and the roughness change of the surrounding rock surface; Step S2: Analyze and process the data information obtained from each sub-area and input it into a pre-learned machine learning model, generate a surrounding rock stress variation coefficient through the machine learning model, and perform an intelligent evaluation of the surrounding rock stress variation based on the surrounding rock variation coefficient; Step S3: Based on the evaluation results of the machine learning model, the surrounding rock stress changes are divided into normal stress changes or abnormal stress changes, and the surrounding rock stress changes are intelligently sensed to obtain stress change evaluation results; Step S4: for the assessed abnormal stress changes, continuously obtain surrounding rock stress change information, and classify the abnormal stress changes into different levels of risk stress changes, including low-risk stress changes, medium-risk stress changes, and high-risk stress changes; Step S5: Select different early warning measures according to the risk stress changes of different levels.

2. According to claim 1, a real-time analysis method for surrounding rock stress distribution in deep complex tunnels is characterized by: The normal direction distribution of the surrounding rock surface is the direction and distribution characteristics of the normal vector of each point on the surrounding rock surface, and the roughness change of the surrounding rock surface is the change in the smoothness or roughness of the surrounding rock surface.

3. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels according to claim 2, characterized in that: The specific steps of analyzing and processing the data information obtained from each sub-region include: The acquired normal direction distribution of the surrounding rock surface and the roughness change of the surrounding rock surface are analyzed to generate the normal direction distribution index and the surface roughness index respectively, and the generated normal direction distribution index and surface roughness index are input into the pre-learned machine learning model.

4. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels according to claim 3, characterized in that: After analyzing the normal direction distribution of the surrounding rock surface, the normal direction distribution index is generated. The specific steps are as follows: Scan and obtain point cloud data of the surrounding rock surface, and calibrate the obtained point cloud data as follows: Among them, P i is the i-th measurement point on the surrounding rock surface, and each point P i All contain three coordinate values, expressed as {x i ,y i 、z i }, x i ,y i 、z i Respectively represent P i The x, y, and z axis coordinates in three-dimensional space, n is the total number of measurement points, that is, the total number of points acquired during the scanning process; By analyzing the neighborhood points around each point, the normal vector of each point is calculated. The calculation expression is as follows: Where N i is the point P on the surrounding rock surface i The normal vector of the surrounding rock surface point P i The neighborhood point set, that is, the point P around the surrounding rock surface i A set of neighboring points, P j is the jth measurement point on the surrounding rock surface, w j is the point P on the surrounding rock surface j The weight coefficient of Calculate each normal vector N i For the angle between the x, y, and z axes, the calculation expression is as follows: In the formula, θ i,x ,θ i,y and θ i,y They are the normal vector N i For the angles of x, y, and z axes; The angle θ of the normal direction i Classify into predefined angle intervals and generate a normal direction histogram. The calculation expression is as follows: In the formula, H x (k), H y (k) and H z (k) are the angular distribution histograms of the normal direction with respect to the x, y, and z axes, respectively, and the angles between all normal vectors and the x, y, and z axes in the angle interval I are counted respectively. k The number of k is the kth angle interval, representing an angle range, used to classify different normal angles into corresponding intervals. as well as middle, is the indicator function, used to determine the angle θ i,x ,θ i,y or θ i,y Is it in interval I k Inside; Angle distribution histogram H for x, y, and z axes x (k), H y (k) and H z (k) Calculate the entropy value and generate the normal direction distribution index. The calculation expression is as follows: Where NDDI is the normal direction distribution index, a∈{x, y, z} is the sum of a, a represents the three coordinate axes x, y, z, m is the total number of angle intervals, H a (k) means that for the coordinate axis a, in the kth angle interval I k The number of normal vectors in .

5. According to claim 3, a real-time analysis method for surrounding rock stress distribution in deep complex tunnels is characterized by: The surface roughness index is generated after analyzing the roughness changes of the surrounding rock surface. The specific steps are as follows: In the detection window, the point cloud data of the surrounding rock surface is obtained from the 3D laser scanner, including each point P i The three-dimensional coordinates x, y, z of each measurement point P are selected i The neighborhood point set is used to calculate the height difference between the measured point and its neighborhood points, and the maximum and minimum height differences are recorded. The calculation expression is as follows: In the formula, ΔH i represents the local height difference of the i-th measuring point, N(i) represents the surrounding rock surface point P i The set of neighboring points of is the height value of the highest point in the neighborhood point set N(i), It is the height value of the lowest point in the neighborhood point set N(i); According to the local height difference, the roughness change rate of each point is calculated and associated with the specific position distance of the detection point i. The calculation expression is as follows: In the formula, R change is the roughness change rate, which indicates the degree of roughness change at the measuring point i, D i is the average neighborhood distance of the measurement point i; The roughness change rate is nonlinearly mapped to generate the roughness index of each point. The calculation expression is as follows: In the formula, R i It's point P i The roughness of the surface, e is the natural base, B is the adjustment parameter used to control the sensitivity of the roughness index to the rate of change, and β is the threshold used to determine the critical value of the roughness change rate; The roughness index of all points is weighted averaged to generate the final surface roughness index. The calculation expression is as follows: Where SRI is the surface roughness index, n is the total number of measurement points, and F i is the weight of measurement point i.

6. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels according to claim 3, characterized in that: The surrounding rock variation coefficient RMVC is generated by the machine learning model, and the formula is as follows: Wherein, k1 and k2 are preset proportional coefficients of the normal direction distribution index NDDI and the surface roughness index SRI, respectively, and both k1 and k2 are greater than 0.

7. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels according to claim 6, characterized in that: The surrounding rock stress change is divided, and the division steps include: If the surrounding rock variation coefficient is greater than or equal to a preset surrounding rock variation coefficient reference threshold, the surrounding rock stress variation is classified as an abnormal stress variation; If the surrounding rock variation coefficient is less than a preset surrounding rock variation coefficient reference threshold, the surrounding rock stress change is classified as a normal stress change.

8. A real-time analysis method for surrounding rock stress distribution in deep complex tunnels according to claim 7, characterized in that: The specific steps of continuously acquiring surrounding rock stress change information for the evaluated abnormal stress change include: By calculating the average value and standard deviation of several surrounding rock variation coefficients in the analysis set, and comparing the obtained average value and standard deviation of surrounding rock variation coefficient with the preset reference threshold value of the average value of surrounding rock variation coefficient and the preset reference threshold value of the standard deviation of surrounding rock variation coefficient, the abnormal stress changes are further divided. The division process is as follows: If the average value of the surrounding rock variation coefficient is greater than or equal to the preset reference threshold value of the surrounding rock variation coefficient, the abnormal stress variation is further classified as a high-risk stress variation; If the average value of the surrounding rock variation coefficient is less than the preset reference threshold value of the surrounding rock variation coefficient and the standard deviation of the surrounding rock variation coefficient is greater than or equal to the preset reference threshold value of the standard deviation of the surrounding rock variation coefficient, the abnormal stress change is further classified as a medium-risk stress change; If the average value of the surrounding rock variation coefficient is less than the preset reference threshold value of the surrounding rock variation coefficient and the standard deviation of the surrounding rock variation coefficient is less than the preset reference threshold value of the standard deviation of the surrounding rock variation coefficient, the abnormal stress change is further classified as a low-risk stress change.

Images