Roadway risk support evaluation method and system based on timing point cloud joint trace evolution

Abstract

A kind of roadway risk support evaluation method and system based on timing point cloud joint trace evolution, method: by laying fixed reference mark point, periodically collect roadway surface three-dimensional point cloud data, and complete multi-period point cloud unified registration and preprocessing;Identify surrounding rock joint trace and extract its geometric and evolution characteristic parameters, realize the cross-period matching of the same joint trace;Divide monitoring subarea and construct joint trace network, based on graph theory analysis and evaluation of surrounding rock risk grade;Combined with the inhibition degree of joint evolution before and after support, evaluate the effectiveness of support, and trigger early warning when relevant characteristic parameters reach threshold, output dangerous site identification, subarea risk result and support optimization suggestion.The system can realize surrounding rock risk identification, support effect evaluation and early warning auxiliary decision-making.The present application can efficiently realize comprehensive evaluation of roadway surrounding rock risk grade and support effectiveness, and can output graded early warning and support suggestion according to different risk grades and support states.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring and safety early warning technology, specifically involving a method and system for assessing roadway risk support based on the evolution of temporal point cloud joint traces. Background Technology

[0002] China possesses abundant mineral resources, but their spatial distribution is significantly uneven due to geological evolution and mineralization conditions. With the continuous depletion of shallow, easily exploitable mineral resources, mineral development is gradually extending to deeper strata, leading to a substantial increase in the proportion of underground mines within the overall mining system. As the core engineering structure of underground mining operations, roadways are crucial channels for personnel passage, equipment transportation, and mineral transshipment; their stability directly determines the safety and continuity of mine production.

[0003] During tunnel excavation and long-term use, the surrounding rock is constantly in a complex mechanical environment, subjected to the coupled effects of multiple factors such as unloading, stress redistribution, and mining disturbances. This makes it prone to the evolution of joints and fissures, including opening, expansion, connection, and even penetration. Once the surrounding rock becomes unstable and fails, it will not only directly threaten the personal safety of underground workers, the integrity of machinery and equipment, and the stability of the tunnel structure itself, but also weaken the control effectiveness of the existing support structure, further exacerbating the risk of disasters such as sidewall spalling, rockfall, and roof collapse. In severe cases, it may lead to mine shutdown and cause significant economic losses.

[0004] Meanwhile, the evolution of joints and fissures in surrounding rock exhibits significant time-varying and spatial variability. This complexity makes it difficult to accurately evaluate the support effect, and the optimization and adjustment of support parameters and the rational determination of support schemes lack scientific basis, hindering timely adaptation. However, current safety management models for surrounding rock in underground mine roadways remain relatively traditional, primarily relying on experience-based support design by engineers, passive reinforcement support, and experience-driven on-site inspections. This leaves surrounding rock control and support management in a reactive state. These management models have significant limitations: they struggle to identify and warn of potential safety risks in roadway surrounding rock in a timely and effective manner, accurately assess the actual effectiveness of existing support structures, and rationally formulate subsequent support optimization schemes based on the evolution of surrounding rock. Consequently, they fail to meet the refined and scientific requirements of safety management for surrounding rock in deep underground mine roadways. Therefore, there is an urgent need to develop a roadway surrounding rock safety management technology that can address these technical challenges. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method and system for assessing roadway risk support based on the evolution of joint traces in time-series point clouds. This method is simple to implement and highly intelligent, efficiently achieving a comprehensive assessment of the risk level of roadway surrounding rock and the effectiveness of support. It can also provide appropriate graded early warnings and support recommendations based on different risk levels and support conditions. The system has a clear structure and high integration, enabling dynamic identification of roadway surrounding rock risks, evaluation of support effectiveness, zonal early warning, and support decision support, significantly improving the safety management level of underground mine roadways.

[0006] To achieve the above objectives, the present invention provides a method for assessing roadway risk support based on temporal point cloud joint trace evolution, comprising the following steps: S1: Multi-phase point cloud data acquisition and preprocessing; Fixed reference markers are set up in the stable area of ​​the roadway, and three-dimensional point cloud data is collected periodically. After registration and preprocessing, a unified spatiotemporal reference is established, and missing areas are marked and supplemented. S2: Joint trace recognition, feature extraction and cross-time matching; Identify the joint traces of surrounding rocks in point clouds of different periods, extract geometric and evolutionary feature parameters, achieve cross-period matching of the same joint traces through comprehensive matching degree, and delay the determination of traces in missing areas; S3: Monitoring zone and joint trace network construction; The tunnel monitoring area is divided into multiple zones, and joint traces are mapped to the corresponding zones. The joint traces are used as nodes and the relationships between traces are used as edges to construct the joint trace network for each zone. S4: Joint network connectivity analysis and surrounding rock risk level assessment; Based on graph theory analysis of the connectivity and evolution characteristics of joint networks in each monitoring zone, dangerous joint types are identified, surrounding rock risk levels are classified according to multi-index thresholds, and adaptive threshold adjustment is supported. S5: Support effectiveness assessment and early warning triggering; By comparing the degree of inhibition of joint trace evolution parameters before and after support, the support inhibition coefficient is calculated, the support effectiveness level is classified, and multi-level early warning and corresponding treatment measures are triggered based on the characteristic parameter threshold. S6: Output of integrated assessment and support optimization recommendations; By integrating the surrounding rock risk level, support effectiveness level, and early warning results, the system outputs hazardous location identification, zoned risk results, and targeted support optimization suggestions.

[0007] This invention provides a method for assessing the risk of surrounding rock and the effectiveness of support in roadways based on the evolution of joint traces in temporal 3D point clouds. First, by setting fixed benchmark points in the stable area of ​​the roadway, a unified spatiotemporal benchmark for multi-period point cloud data is established, solving the key problem of inconsistent coordinates across different periods. Periodic acquisition combined with registration preprocessing achieves point cloud denoising, anomaly removal, and missing area marking, ensuring data quality and continuity. Marking missing areas and generating supplementary prompts effectively avoids misjudgments due to data gaps, laying a reliable data foundation for accurate identification of subsequent joint traces. Second, continuous identification of exposed structural surfaces on the roadway surface helps to accurately extract multi-dimensional features, including spatial location, length, orientation, opening degree, and various temporal evolution parameters. By constructing a comprehensive matching degree model, accurate positioning and tracking of the same joint trace in point clouds at different periods are achieved, solving the problem of cross-period data correlation. A delayed judgment mechanism is adopted for traces in missing areas, enhancing the robustness of the method under complex working conditions. Next, the tunnel monitoring area was scientifically divided into multiple functional zones, and joint traces were mapped to the corresponding zones according to their spatial location. A zoned joint trace network was constructed using joint traces as nodes, geometric intersections between traces, endpoint proximity, and reachability as edges. This transformed the complex joint distribution relationships into a quantifiable graph structure, providing a clear mathematical foundation for subsequent connectivity analysis. Subsequently, based on graph theory methods, key indicators such as the scale of connected components, node degree, network connectivity, and penetration rate of the joint network in each zone were analyzed, achieving a quantitative characterization of the connectivity and evolution characteristics of the joint network. It can identify various hazardous joint types, including open, extended, and connected joints, and classify the surrounding rock risk into five levels based on multi-index thresholds. Simultaneously, it supports adaptive adjustment of thresholds based on historical monitoring data, balancing the stability of the assessment criteria with the dynamic adaptability to on-site conditions. Then, by comparing the changes in evolution parameters of the same joint trace before and after support, the support inhibition coefficient is calculated to quantitatively evaluate the inhibition effect of support on joint expansion, opening, and network connectivity. Based on the degree of inhibition, the support effectiveness is divided into five levels. Furthermore, based on different threshold combinations of characteristic parameters such as opening rate increase, length rate increase, and network connectivity, multi-level early warning signals can be triggered, and corresponding graded treatment measures can be output, achieving closed-loop management from assessment to early warning. Finally, by integrating the surrounding rock risk level, support effectiveness level, and early warning results, dangerous areas of the roadway surrounding rock are accurately identified, and a zonal risk result map is generated. Based on different combinations of risk and effectiveness levels, targeted support optimization suggestions can be output, such as maintenance monitoring, local reinforcement, increasing the support level, or strengthening support in key areas, providing direct and operable technical basis for on-site decision-making.This invention can continuously identify, match across time periods and analyze the evolution of joint traces in roadway surrounding rock. It can achieve fully automated processing of the entire process, including identification of roadway surrounding rock joint evolution risks, dynamic evaluation of support effects, zonal early warning and output of support optimization suggestions. It can provide an integrated technical solution for underground mine roadway safety, from data acquisition, analysis and evaluation to decision support.

[0008] This method is simple to implement and highly intelligent. Through automatic point cloud acquisition and registration, intelligent joint trace identification and matching, graph theory network connectivity analysis, and quantitative calculation of support inhibition coefficient, it can efficiently achieve a comprehensive assessment of the risk level of the surrounding rock and the effectiveness of the support in roadways. It can also output appropriate graded early warnings and support suggestions based on different risk levels and support conditions. It has significant technological advancements and practical engineering value, avoiding the problems of strong subjectivity, low efficiency, and difficulty in quantification in traditional manual inspection methods. It is of great significance for ensuring safe production operations in underground mine roadways.

[0009] Furthermore, to ensure the integrity and comparability of time-series point cloud data, the multi-period point cloud data acquisition and preprocessing process in S1 is as follows: S11: Set a reference point; Set a fixed reference point in a location where the surrounding rock is stable and unaffected by construction disturbance, as a unified spatial coordinate reference for multi-phase point clouds; S12: Collect point cloud data; periodically collect 3D point cloud data of the tunnel surface; S13: Multi-stage point cloud registration; Initial registration is performed first based on fixed reference markers, and then fine registration is performed based on overlapping areas of the point clouds; S14: Point cloud preprocessing; denoising, outlier removal, monitoring area cropping, coordinate standardization, and missing area marking of 3D point cloud data; S15: Missing area handling; When a point cloud in a certain period is missing due to occlusion, support structure coverage, or data acquisition reasons, it is marked as a temporarily missing area, and a prompt message for manual supplementary scanning is generated.

[0010] In this technical solution, a unified spatial coordinate benchmark for multi-period point cloud data is established by setting up fixed benchmark points in the stable area of ​​the surrounding rock, fundamentally solving the key problem of inconsistent coordinates among multi-period data. A two-level registration strategy based on the benchmark points and on the overlapping areas is adopted, which significantly improves the registration accuracy while ensuring registration efficiency. The system's preprocessing process includes noise reduction, anomaly removal, cropping, standardization, and missing data marking, which comprehensively improves the quality of point cloud data. In particular, for data missing areas caused by occlusion, support coverage, etc., the system is clearly marked as temporarily missing areas and generates manual supplementary scanning prompts, avoiding misjudgments or omissions caused by data missingness, and laying a reliable data foundation for the accurate identification and cross-period matching of subsequent joint traces.

[0011] Furthermore, in order to achieve accurate identification and automatic cross-time tracking of joint traces, and to enhance the robustness of matching under data missing conditions by combining a delayed decision mechanism, the process of joint trace identification, feature extraction, and cross-time matching in S2 is as follows: S21: Identify joint traces; Identify surrounding rock joint traces from point clouds of different phases; S22: Extract geometric and evolutionary characteristic parameters of joint traces; extract spatial location, endpoint coordinates, length, orientation, and opening characteristic parameters, as well as temporal evolution parameters such as length change, length growth rate, opening change, opening growth rate, and orientation change. The opening feature parameters are obtained by the normal distance between the point cloud boundaries on both sides of the joint trace. When they cannot be obtained stably, the apparent opening width is calculated by using local geometric concavity features and the distance between the crack edges, or an equivalent opening parameter is constructed as a substitute. Among them, the first is obtained according to formula (1). Opening characteristic parameters of joint traces ; (1); In the formula, The distance between the two boundary point clouds extracted from the profile along the joint trace direction; The number of effective profiles used in the calculation; S23: Cross-period matching determination; setting a comprehensive matching degree threshold. Based on four dimensions—spatial proximity, directional consistency, endpoint correspondence, and extension continuity—normalized indices are calculated for each dimension, and then weighted and summed to obtain the comprehensive matching degree. ;when At that time, it was determined that the joint lines from two different periods met the cross-period matching condition; Among them, according to formula (2) Period Joint traces and Issue No. Joint trace line comprehensive matching degree ; (2); In the formula, As a spatial proximity normalization index, As a directional consistency normalization index, As a normalized index for endpoint correspondence, To extend the continuous normalization index; , , , These are the weights of the spatial proximity normalization index, the directional consistency normalization index, the endpoint correspondence normalization index, and the extension continuity normalization index, respectively. ; S24: Delayed determination; when a joint trace that existed in the previous period is not identified in the later period and the corresponding area is marked as a temporarily missing area, the joint trace is marked as pending confirmation and a delayed determination is made in conjunction with subsequent periodic data.

[0012] In this technical solution, firstly, in the feature extraction stage, not only are static geometric parameters such as the spatial location, length, and orientation of joint traces extracted, but also temporal evolution parameters such as length change and opening growth rate are extracted. Furthermore, opening characteristic parameters are obtained through multi-point sampling along the trace direction, providing rich quantitative indicators for subsequent dynamic risk assessment. Secondly, in the cross-period matching stage, a comprehensive matching degree model is constructed, encompassing four dimensions: spatial proximity, directional consistency, endpoint correspondence, and extension continuity. Through weighted summation, qualitative matching principles are transformed into a quantitatively calculable comprehensive matching degree index. When the matching degree reaches a preset threshold, it is automatically determined to be the same joint, achieving efficient automatic association of joint traces from different periods. In addition, for unmatched traces due to occlusion or data loss, a delayed judgment mechanism is adopted, marking them as pending confirmation and continuing tracking based on subsequent periodic data. This avoids misjudgments or omissions caused by missing data in a single period, significantly improving the method's adaptability and matching reliability in complex mining environments.

[0013] Furthermore, in order to transform the complex joint distribution into a graph structure that can be quantified and analyzed, the process of constructing the monitoring zone and joint trace network in S3 is as follows: S31: Divide the monitoring area into six zones: the arch zone, the left shoulder zone, the right shoulder zone, the left side zone, the right side zone, and the floor zone. S32: Trace mapping: Based on the spatial location, endpoint coordinates and area range of the joint trace, the joint trace is mapped to the corresponding monitoring zone; when the joint trace crosses two or more monitoring zones, the monitoring zone to which it belongs is determined according to the area where its main body is located or the area with the largest length proportion. S33: Construct a network; using the joint traces in each monitoring zone as network nodes, establish connection edges between nodes based on the intersection relationship, endpoint proximity relationship, or extension reachability relationship of the traces to form a joint trace network for the corresponding monitoring zone.

[0014] In this technical solution, firstly, the roadway monitoring area is divided into six zones: the arch zone, the left shoulder zone, the right shoulder zone, the left side zone, the right side zone, and the floor zone. This division method highly matches the roadway's stress and deformation characteristics and risk distribution patterns, making the subsequent risk assessment results more engineering-oriented and practically instructive. Secondly, in the trace mapping stage, joint traces are assigned to corresponding zones based on their spatial location and endpoint coordinates. Traces spanning multiple zones are assigned based on the principle of having the largest proportion of the main area or length. The processing rules are clear and highly operable, avoiding duplicate calculations or omissions caused by unclear trace assignments. Finally, in the network construction stage, joint traces within each zone are used as network nodes. Connection edges between nodes are established based on trace intersection relationships, endpoint proximity relationships, and extension reachability relationships. This transforms the complex spatial distribution relationship of joints into a clear and quantifiable graph theory network structure, laying a solid mathematical model foundation for subsequent quantitative analysis based on graph theory, such as the scale of connected components, node degree, network connectivity, and penetration rate.

[0015] Furthermore, in order to achieve a five-level quantitative assessment of surrounding rock risk, the process of joint network connectivity analysis and surrounding rock risk level assessment in S4 is as follows: S41: Calculate network connectivity characteristics: Calculate the scale of connected components, node degree, network connectivity, percentage of the largest connected subgraph, key connecting paths and connectivity rate of each monitoring zone to characterize the connectivity and evolution characteristics of the joint network. Among them, network connectivity is calculated according to formula (3). ; (3); In the formula, The number of nodes in the partitioned joint trace network. This represents the number of network edges. The penetration rate is calculated according to formula (4). ; (4); In the formula, To determine the number of joint clusters required to form a critical through path, This represents the total number of joint clusters in the partition; S42: Identify hazardous joint types; based on opening characteristics, length change, length growth rate, opening degree change, opening degree growth rate, connectivity characteristics, and location characteristics, identify hazardous joint types; the hazardous joint types include opening type, extension type, connectivity type, through type, activation type, and hazardous joints in critical locations; S43: Classify the risk level of the surrounding rock; classify the risk level of the surrounding rock in each monitoring zone according to the geometric and evolutionary characteristic parameters of joint traces and the connectivity and evolution characteristics of joint networks; the risk level of the surrounding rock includes five levels: safe, relatively safe, relatively dangerous, dangerous, and high dangerous.

[0016] In this technical solution, firstly, in the network connectivity characteristic calculation stage, two quantitative indicators, network connectivity degree and penetration rate, are introduced. Network connectivity degree describes the overall connection density of the network through the relationship between the number of nodes and the number of edges, while penetration rate reflects the degree of network penetration by the ratio of the number of joint clusters in key penetration paths to the total number of joint clusters. The two indicators complement each other, achieving a comprehensive quantitative characterization of the connectivity status of the joint network. Secondly, in the dangerous joint type identification stage, multiple dimensions such as opening characteristics, length change, length growth rate, opening degree change, opening degree growth rate, connectivity characteristics, and location characteristics are comprehensively considered. This allows for the identification of six types: opening type, extended type, connected type, penetrating type, activated type, and dangerous joints in key locations. The detailed and comprehensive classification provides a clear basis for risk tracing and targeted support. Finally, in the risk level classification stage, the surrounding rock risk is divided into five levels: safe, relatively safe, relatively dangerous, dangerous, and high dangerous. The granularity is reasonable and facilitates on-site hierarchical management and differentiated response, providing a clear risk benchmark for subsequent early warning triggering and support optimization suggestions.

[0017] Furthermore, to achieve dynamic optimization of the assessment criteria and flexible, manually controllable management, in S43, the surrounding rock risk level is determined based on the threshold ranges of the opening growth rate, length growth rate, network connectivity, and penetration rate. Each indicator's threshold range is initially defined using preset fixed thresholds and adaptively adjusted based on historical monitoring data, regional noise levels, and changes in support response. The adaptive adjustment range is limited to a preset allowable range; when it exceeds this range, it takes effect only after manual authorization and confirmation. This mechanism, combining multi-indicator comprehensive judgment with adaptive threshold adjustment, ensures both the initial consistency and stability of the assessment criteria and dynamically adapts to changes in different working conditions. Simultaneously, the dual control of range limitation and manual authorization ensures the safety and controllability of the adjustment process.

[0018] Furthermore, in order to achieve closed-loop management of support effectiveness level classification and multi-level risk response, the support effectiveness assessment and early warning triggering process in S5 is as follows: S51: Calculate the support inhibition coefficient; compare the changes in joint trace length, length growth rate, opening change, opening growth rate, network connectivity change rate, and changes in the development state of dangerous joints before and after support, and calculate the support inhibition coefficient. ; through support inhibition coefficient Quantitatively assess the degree to which support inhibits joint trace expansion, opening, and network connectivity evolution; if the change before and after support is zero, then... If the change before support is zero but the change after support is greater than zero, then ; The support inhibition coefficient is calculated according to formula (5). ; (5); In the formula, , , These represent the changes in opening, length, and network connectivity before support was installed. , , These represent the changes in opening, length, and network connectivity after support. , , The weights corresponding to the changes in opening, length, and connectivity after support are given, and ; S52: Classification of Support Effectiveness Levels: Based on the support inhibition coefficient and the development status of dangerous joints, the support effectiveness levels are classified; the support effectiveness levels include five levels: significantly effective, effective, average, poor, and ineffective; S53: Triggering an early warning signal; triggering an early warning and corresponding handling measures based on relevant characteristic parameters and support effectiveness assessment results: When the opening growth rate exceeds the first preset threshold B, a level one early warning is triggered, and measures to strengthen inspection, continuous monitoring, and key retesting of abnormal zones are output. When the growth rate of crack length and network connectivity both exceed the second preset threshold B, a level-two early warning is triggered, and measures such as special investigation of local areas, addition of monitoring points, and temporary reinforcement are output. When a critical breakthrough path is formed or the breakthrough rate exceeds the third preset threshold B, a level three early warning is triggered, and measures to strengthen support in key areas and restrict operations in related areas are output. When the dangerous joint continues to develop after support, or when the support effectiveness is lower than the fourth preset threshold B, an upgraded warning is triggered, and measures such as encrypted monitoring, enhanced on-site control, and, if necessary, the evacuation of personnel from the dangerous area are output.

[0019] In this technical solution, firstly, in the calculation of the support inhibition coefficient, a weighted comprehensive formula for the support inhibition coefficient is constructed by comparing the changes in joint opening, length, and network connectivity before and after support. This quantifies the inhibition effect of support on joint expansion, opening, and network connectivity evolution into a single numerical index. A specific processing rule is set when the change before support is zero, avoiding calculation anomalies and ensuring the robustness of the evaluation method. Secondly, in the classification of support effectiveness levels, effectiveness is divided into five levels—significantly effective, effective, average, poor, and ineffective—based on the magnitude of the support inhibition coefficient, providing fine granularity. The system is rationally designed to facilitate hierarchical management on-site. Furthermore, the threshold levels can be adaptively adjusted based on historical monitoring results and supplemented by manual authorization confirmation, balancing the stability and dynamic adaptability of the assessment standards. Finally, a four-level early warning mechanism has been established at the early warning triggering stage, corresponding to different risk scenarios such as exceeding the standard for opening growth rate, exceeding the standard for both length growth rate and connectivity, formation of critical through-path, and continued development of dangerous joints after support. Each level of early warning is equipped with clear handling measures, achieving closed-loop management of the entire process from monitoring and assessment to hierarchical response, providing clear and operable technical basis for on-site safety decisions.

[0020] Furthermore, to ensure both stability and flexibility in the level determination, in S52, the support effectiveness level is determined based on the level threshold range of the support inhibition coefficient. This level threshold range is initially divided using a preset fixed threshold and is adaptively adjusted based on historical monitoring results. The adaptive adjustment range is limited to a preset allowable range; when it exceeds this range, it takes effect only after manual authorization and confirmation. This mechanism, through a two-tiered strategy of preset fixed thresholds and adaptive adjustment, ensures both the initial consistency and operability of the support effectiveness level determination, and dynamically optimizes the threshold based on historical monitoring results to adapt to different working conditions. Simultaneously, the dual control of range limitation and manual authorization ensures the safety and controllability of the adjustment process.

[0021] Furthermore, in order to achieve a closed-loop output from assessment to decision-making, the process of integrating assessment and support optimization recommendations in S6 is as follows: S61: Integrated assessment; A comprehensive assessment is conducted by taking into account the surrounding rock risk level, support effectiveness level, and early warning results; S62: Output support optimization recommendations; Output support optimization recommendations based on overall assessment results: When the surrounding rock risk level is safe or relatively safe, and the support effectiveness level is significantly effective or effective, the recommendation is to maintain the existing support and continue monitoring. When a local joint continues to expand or connectivity increases but no critical through path is formed, a local reinforcement and support suggestion is output. When the surrounding rock risk level is relatively dangerous, dangerous or high dangerous, and the support effectiveness level is average, poor or ineffective, a recommendation to increase the support level is issued. When a critical connection path is formed or dangerous joints remain active after support, recommendations are made to strengthen support and increase monitoring in key areas. S63: Output identification and results; Based on the surrounding rock risk level, dangerous joint type, distribution of key penetration paths and support effectiveness level of each monitoring zone, identify dangerous parts of the surrounding rock in the roadway; generate a zone risk result map to visually display the risk level, dangerous joint distribution and key penetration paths of each zone; output dangerous part identification, zone risk results and corresponding support optimization suggestions.

[0022] In this technical solution, firstly, in the integrated assessment stage, information from three dimensions—surrounding rock risk level, support effectiveness level, and early warning results—is comprehensively evaluated to avoid the one-sidedness of single-indicator evaluation and provide a comprehensive and reliable basis for subsequent decision-making. Secondly, in the support optimization suggestion output stage, differentiated suggestions are output for four typical scenarios: maintaining monitoring under safe conditions, local reinforcement during local expansion, increasing the support level when there is high risk and support failure, and strengthening support and increasing monitoring when a through path is formed. The suggestions are comprehensive and highly targeted, providing clear and actionable technical guidance for on-site personnel. Finally, in the output identification and results stage, dangerous areas are accurately identified to form a zonal risk result map, visually displaying the risk level, dangerous joint distribution, and key through paths of each zone. This transforms abstract risk data into intuitive graphical information, facilitating on-site personnel to quickly locate risk areas, understand the risk situation, and take corresponding measures, achieving a closed-loop process from data collection, analysis and evaluation to visual decision support.

[0023] This invention also provides a roadway risk support assessment system based on temporal point cloud joint trace evolution, used to implement a roadway risk support assessment method based on temporal point cloud joint trace evolution, including: The point cloud acquisition module is used to periodically acquire three-dimensional point cloud data of the roadway surface in the roadway monitoring area, and at the same time record the spatial coordinate information of fixed reference marker points. A unified registration module is used to achieve a unified spatial coordinate reference for point clouds from multiple periods and to preprocess the registered point cloud data; The joint trace identification module is used to identify the joint traces of the surrounding rock from the point cloud of each phase and extract the geometric and evolutionary feature parameters of the joint traces. A cross-period matching module is used to achieve cross-period matching of the same joint trace based on spatial proximity, directional consistency, endpoint correspondence and extension continuity. The partition modeling module is used to divide the roadway monitoring area into six monitoring partitions: the arch area, the left shoulder area, the right shoulder area, the left side area, the right side area, and the floor area, and to map joint traces to the corresponding monitoring partitions; at the same time, it is used to establish the joint trace network of the corresponding monitoring partitions. The network connectivity analysis module is used to analyze the connectivity and evolution characteristics of the joint trace network in each monitoring zone, identify dangerous joint types, and assess the surrounding rock risk level of each monitoring zone. The support effectiveness assessment module is used to quantitatively assess the degree of inhibition of the expansion, opening and network connectivity evolution of the joint trace by comparing the changes in evolution parameters of the same joint trace before and after support implementation, and to determine the support effectiveness level based on the degree of inhibition of joint evolution and the development status of dangerous joints. The early warning and support suggestion output module is used to trigger the corresponding early warning signal when the relevant characteristic parameters reach the preset threshold, and output the dangerous part identification, zoning risk results and support optimization suggestions according to the surrounding rock risk level, support effectiveness level and early warning results.

[0024] This invention provides a roadway risk support assessment system based on the evolution of joint traces in time-series point clouds. It integrates point cloud acquisition, unified registration, joint trace identification, network connectivity analysis, risk assessment, support effectiveness evaluation, early warning, and support suggestion output. Through the point cloud acquisition module, a periodic acquisition strategy ensures the temporal comparability of the data. Simultaneously, the synchronous recording of marker point coordinates provides a reliable basis for subsequent high-precision registration. The unified registration module effectively eliminates systematic biases between data collected from different periods, ensuring the accuracy of subsequent joint trace comparison analysis and significantly improving data quality, laying a data foundation for reliable joint trace identification. Furthermore, the collaborative work of the point cloud acquisition module and the unified registration module effectively establishes a unified spatiotemporal benchmark for multi-period point clouds while ensuring data quality. The joint trace identification module automatically extracts the spatial location, endpoint coordinates, length, direction, opening characteristic parameters, and temporal evolution parameters such as length change and opening rate of joint traces from massive point cloud data, achieving a structured expression of joint information. By implementing a cross-period matching module, automatic association and tracking of joint traces from different periods are achieved. A delayed judgment mechanism is employed for traces in missing areas, enhancing the method's adaptability under incomplete data conditions. Simultaneously, the linkage between the joint trace identification module and the cross-period matching module enables continuous tracking and evolution feature extraction of joint traces. The partitioning modeling module scientifically matches the partitioning method with the roadway's stress characteristics and risk distribution patterns, making the risk assessment results more engineering-oriented. It also transforms complex joint distributions into quantifiable graph structures, providing a clear mathematical model for subsequent graph theory analysis. The network connectivity analysis module accurately characterizes joint network connectivity features, identifying various hazardous joint types. The surrounding rock risk level classification balances the stability of the assessment criteria with the dynamic adaptability to on-site conditions. Furthermore, the combination of the partitioning modeling module and the network connectivity analysis module transforms complex joint distributions into quantifiable graph networks, enabling accurate and efficient assessment of the surrounding rock risk level. By setting up the support effectiveness assessment module, the support effect can be quantitatively characterized through the support inhibition coefficient, transforming qualitative evaluation into a comparable numerical indicator. Furthermore, the finer granularity of the support effectiveness level facilitates subsequent on-site hierarchical management. The early warning and support recommendation output module enables dynamic risk response based on a multi-level early warning mechanism, simultaneously obtaining targeted support recommendations.

[0025] The system has a clear structure and high integration, enabling dynamic identification of roadway surrounding rock risks, evaluation of support effectiveness, zonal early warning, and support decision support. It significantly improves the level of safety management of surrounding rock in underground mine roadways and effectively avoids the problems of strong subjectivity, low efficiency, and difficulty in quantification in traditional manual inspection methods. It has significant technological advancement and engineering practical value for ensuring safe production operations in underground mine roadways and improving the level of surrounding rock safety management. Attached Figure Description

[0026] Figure 1 This is a flowchart of the evaluation method in this invention; Figure 2 This is a schematic diagram of the tunnel point cloud acquisition and structural surface information extraction in this invention; Figure 3 This is a schematic diagram of the structural surface recognition results in this invention; Figure 4 This is a schematic diagram of the roadway monitoring zones in this invention; Figure 5 This is a block diagram illustrating the principle of the evaluation system in this invention; In the diagram: 201, roof; 202, roof control point; 203, roof joint traces; 204, roadway side joint traces; 205, upper roadway side control point; 206, point cloud; 207, 3D laser scanner; 208, middle roadway side control point; 209, mine car; 210, lower roadway side control point; 211, floor; 212, mine car track. Detailed Implementation

[0027] The invention will now be further described with reference to the accompanying drawings.

[0028] like Figure 1 As shown, this invention provides a method for assessing roadway risk support based on temporal point cloud joint trace evolution, comprising the following steps: S1: Multi-phase point cloud data acquisition and preprocessing; Fixed reference markers are set up in the stable area of ​​the roadway, and three-dimensional point cloud data is collected periodically. After registration and preprocessing, a unified spatiotemporal reference is established, and missing areas are marked and supplemented. like Figure 2 As shown, Figure 2The diagram illustrates the point cloud acquisition and structural surface information extraction of the roadway according to the present invention, including roof 201, roof control points 202, roof joint traces 203, roadway side joint traces 204, upper roadway side control points 205, point cloud 206, 3D laser scanner 207, middle roadway side control points 208, mine car 209, lower roadway side control points 210, floor 211, and mine car track 212. Among them, the mine car 209 is set on the mine car track 212, and the three-dimensional laser scanner 207 is installed above the mine car 209 for periodic moving scans along the roof 201; the roof control point 202 is set in a position where the surrounding rock of the roadway is relatively stable and not easily affected by construction disturbance, and is used to establish a unified spatial coordinate benchmark for multi-phase point clouds; the three-dimensional laser scanner 207 scans the roadway roof control point 202, roof joint traces 203, lower control point 210 of the roadway sidewall and floor 211 through scanning beam lines to obtain the roadway surface point cloud 206, and extract the surrounding rock structural surface information based on the collected point cloud data.

[0029] To ensure the integrity and comparability of time-series point cloud data, the multi-period point cloud data acquisition and preprocessing process is as follows: S11: Set a reference point; Set a fixed reference point in a location where the surrounding rock is stable and unaffected by construction disturbance, as a unified spatial coordinate reference for multi-phase point clouds; As a preferred embodiment, the fixed reference markers are preferably located in relatively stable areas on both sides and the roof of the tunnel, corresponding to... Figure 2 The control points 202 (top plate), 205 (upper side of roadway), 208 (middle side of roadway), and 210 (lower side of roadway) are used to improve the registration accuracy of multi-phase point clouds and establish a unified spatial coordinate benchmark.

[0030] S12: Collect point cloud data; periodically collect 3D point cloud data of the tunnel surface; The three-dimensional point cloud acquisition cycle is set according to the stability of the surrounding rock in the roadway, preferably once a week or immediately after blasting and after support. For stable areas, a regular cycle acquisition method is used. For areas with strong surrounding rock disturbance, blasting areas, or support areas, a short cycle and dense acquisition method is used. For stable areas, a regular cycle acquisition method is used.

[0031] As a preferred method, a three-dimensional laser scanner 207 installed on the mine car 209 is used to scan the roadway monitoring area along the mine car track 212 according to a preset acquisition cycle to form a multi-period point cloud raw dataset.

[0032] S13: Multi-phase point cloud registration; To ensure high quality of the acquired point cloud data, after the original point cloud acquisition is completed, initial registration is first performed based on fixed reference marker points (roof control point 202, upper sidewall control point 205, middle sidewall control point 208, lower sidewall control point 210), and then fine registration is performed based on the overlapping areas between adjacent point clouds, thereby achieving unified coordinate alignment of multi-phase point clouds; Preferably, the average registration error after multi-stage fine registration of point clouds is no greater than 5mm, and the effective coverage rate of the point cloud after preprocessing is no less than 90%. S14: Point cloud preprocessing; after registration, the 3D point cloud data is denoised, outlier removal, monitoring area cropping, coordinate standardization, and missing area marking; wherein, the point cloud denoising preferably adopts neighborhood filtering or statistical filtering, and outlier removal is preferably determined based on distance threshold and neighborhood point density; the effective point cloud coverage after preprocessing is preferably not less than 90%.

[0033] S15: Missing area handling; When a point cloud in a certain period is missing due to occlusion, support structure coverage, or data acquisition reasons, it is marked as a temporarily missing area, and a prompt message for manual supplementary scanning is generated.

[0034] Preferably, when the missing area of ​​the point cloud in a local area accounts for more than 10% of the area of ​​the corresponding monitoring zone, the area is marked as a temporarily missing area; when the same area is marked as a temporarily missing area for two consecutive periods, a manual supplementary scan prompt message is generated.

[0035] Preferably, when the missing area in a local area accounts for more than 10% of the area of ​​the corresponding monitoring zone, the area is marked as a temporarily missing area; when the same area is marked as a temporarily missing area for two consecutive periods, a manual supplementary scan or on-site verification prompt is triggered. Through the above steps, high-precision and time-consistent three-dimensional point cloud data of the roadway surface can be obtained, providing a reliable data foundation for subsequent joint trace identification, cross-period matching, and zone risk assessment.

[0036] As a preferred embodiment, the main parameter settings in S1 are shown in Table 1: Table 1: Point Cloud Acquisition and Preprocessing Parameter Settings In this technical solution, a unified spatial coordinate benchmark for multi-period point cloud data is established by setting up fixed benchmark points in the stable area of ​​the surrounding rock, fundamentally solving the key problem of inconsistent coordinates among multi-period data. A two-level registration strategy based on the benchmark points and on the overlapping areas is adopted, which significantly improves the registration accuracy while ensuring registration efficiency. The system's preprocessing process includes noise reduction, anomaly removal, cropping, standardization, and missing data marking, which comprehensively improves the quality of point cloud data. In particular, for data missing areas caused by occlusion, support coverage, etc., the system is clearly marked as temporarily missing areas and generates manual supplementary scanning prompts, avoiding misjudgments or omissions caused by data missingness, and laying a reliable data foundation for the accurate identification and cross-period matching of subsequent joint traces.

[0037] S2: Joint trace recognition, feature extraction and cross-time matching; Identify the joint traces of surrounding rocks in point clouds of different periods, extract geometric and evolutionary feature parameters, achieve cross-period matching of the same joint traces through comprehensive matching degree, and delay the determination of traces in missing areas; To achieve accurate identification and automatic cross-temporal tracking of joint traces, and to enhance matching robustness under data missing conditions by incorporating a delayed decision mechanism, the process of joint trace identification, feature extraction, and cross-temporal matching is as follows: S21: Identify joint traces; identify the joint traces of the surrounding rock from point clouds of different phases; such as Figure 3 As shown, Figure 3 A schematic diagram showing the structural surface recognition results of the present invention is presented.

[0038] In a preferred embodiment, the exposed structural surfaces of the roadway are first identified based on the standard point cloud data obtained in S1, and the boundary segments of the structural surfaces are extracted to form a set of joint traces. The joint traces are preferably expressed in the form of line segments, and each joint trace includes at least the coordinates of the starting point, the ending point, the midpoint, the trace length, and the orientation angle. Further, the trace length can be calculated from the Euclidean distance between the starting and ending points, and the trace orientation can be determined by the projection direction of the joint trace in the local coordinate system of the roadway.

[0039] S22: Extract geometric and evolutionary characteristic parameters of joint traces; extract spatial location, endpoint coordinates, length, orientation, opening characteristic parameters, as well as temporal evolution parameters such as length change, length growth rate, opening change, opening growth rate, and orientation change as geometric and evolutionary characteristic parameters of joint traces; The opening feature parameters are preferentially obtained by the normal distance between the point cloud boundaries on both sides of the joint trace. When they cannot be obtained stably, the apparent opening width is calculated by using local geometric concavity features and the distance between the crack edges, or an equivalent opening parameter is constructed as a substitute. Among them, the first is obtained according to formula (1). Opening characteristic parameters of joint traces ; (1); In the formula, The distance between the two boundary point clouds extracted from the profile along the joint trace direction; The number of effective profiles used in the calculation; As a preferred embodiment, the length change is the difference in length of the same joint trace in two adjacent periods, the opening change is the difference in opening characteristic parameters of the same joint trace in two adjacent periods, and the azimuth change is the difference in azimuth angle of the same joint trace in two adjacent periods. The length growth rate... and opening growth rate The values ​​are determined by dividing the changes in joint trace length and opening between two adjacent periods by the corresponding parameter value of the previous period. Preferably, when the corresponding parameter value of the previous period is less than a preset minimum reference value, the minimum reference value is used as a substitute to avoid abnormal calculations caused by an excessively small denominator. Preferably, the minimum reference value is 1 mm.

[0040] S23: Cross-period matching determination; Cross-period matching of the same joint trace is jointly determined based on spatial proximity, directional consistency, endpoint correspondence, and extension continuity. Specifically, a comprehensive matching degree threshold is set. Based on four dimensions—spatial proximity, directional consistency, endpoint correspondence, and extension continuity—normalized indices are calculated for each dimension, and then weighted and summed to obtain the comprehensive matching degree. ;when When two joint lines from different periods meet the cross-period matching condition, they are considered to be the same joint. Among them, according to formula (2) Period Joint traces and Issue No. Joint trace line comprehensive matching degree ; (2); In the formula, As a spatial proximity normalization index, As a directional consistency normalization index, As a normalized index for endpoint correspondence, To extend the continuous normalization index, and the larger the value of each normalization index, the higher the degree of matching; , , , These are the weights of the spatial proximity normalization index, the directional consistency normalization index, the endpoint correspondence normalization index, and the extension continuity normalization index, respectively. ; As a preferred option, the first preset threshold A is set to 5mm to 30mm, preferably 10mm under conventional roadway monitoring conditions; the second preset threshold A is set to 5° to 20°, preferably 10° under conventional roadway monitoring conditions; the third preset threshold A is set to 5mm to 25mm, preferably 10mm under conventional roadway monitoring conditions; and a comprehensive matching degree threshold is set. The value is 0.70–0.90, and 0.80 is preferred under conventional roadway monitoring conditions; The spatial proximity normalization index is obtained as follows: the distance between the midpoints or centroids of two adjacent joint traces is calculated, and this distance is compared with a first preset threshold A. If the distance is zero, the index takes the maximum value; if the distance reaches or exceeds the first preset threshold A, the index takes the minimum value of zero; if the distance is between the two, the index is calculated in a linear decay manner, with the index decreasing as the distance increases.

[0041] The directional consistency normalization index is obtained as follows: Calculate the angle between the directions of two adjacent joint traces (taking an acute angle, ranging from 0° to 90°), and compare this angle with a second preset threshold A. If the angle is zero, the index takes its maximum value; if the angle reaches or exceeds the second preset threshold A, the index takes its minimum value of zero; if the angle is between the two, the index is calculated in a linear decay manner, with the larger the angle, the smaller the index.

[0042] The endpoint correspondence normalization index is obtained as follows: the distance between the corresponding endpoints of two sets of joint traces in two adjacent periods is calculated, and the smaller value is taken as the endpoint distance. This distance is then compared with a third preset threshold A. If the endpoint distance is zero, the index is set to its maximum value; if the endpoint distance reaches or exceeds the third preset threshold A, the index is set to its minimum value of zero; if the endpoint distance is between the two, the index is calculated in a linear decay manner, with the index decreasing as the distance increases.

[0043] The extension continuity normalization index is obtained as follows: This index is a binary index. It determines whether the endpoint of the joint trace in the later period is located near the joint trace in the previous period or within the allowable range of its extension direction. Specifically, it calculates the vertical distance from the endpoint of the joint trace in the later period to the joint trace in the previous period or its extension direction line. If this vertical distance is less than a set allowable range (e.g., 10mm), the index value is 1; otherwise, it is 0.

[0044] As a preferred implementation, the cross-period matching threshold parameter settings in S2 are shown in Table 2: Table 2: Cross-Period Matching Threshold Parameter Table S24: Delayed determination; when a joint trace that existed in the previous period is not identified in the later period and the corresponding area is marked as a temporarily missing area, the joint trace is marked as pending confirmation and a delayed determination is made in combination with subsequent period data; as an preferred option, when the same joint trace is in pending confirmation for two consecutive periods, a manual review or supplementary scan prompt message is generated.

[0045] In this technical solution, firstly, in the feature extraction stage, not only are static geometric parameters such as the spatial location, length, and orientation of joint traces extracted, but also temporal evolution parameters such as length change and opening growth rate are extracted. Furthermore, opening characteristic parameters are obtained through multi-point sampling along the trace direction, providing rich quantitative indicators for subsequent dynamic risk assessment. Secondly, in the cross-period matching stage, a comprehensive matching degree model is constructed, encompassing four dimensions: spatial proximity, directional consistency, endpoint correspondence, and extension continuity. Through weighted summation, qualitative matching principles are transformed into a quantitatively calculable comprehensive matching degree index. When the matching degree reaches a preset threshold, it is automatically determined to be the same joint, achieving efficient automatic association of joint traces from different periods. In addition, for unmatched traces due to occlusion or data loss, a delayed judgment mechanism is adopted, marking them as pending confirmation and continuing tracking based on subsequent periodic data. This avoids misjudgments or omissions caused by missing data in a single period, significantly improving the method's adaptability and matching reliability in complex mining environments.

[0046] S3: Monitoring zone and joint trace network construction; The tunnel monitoring area is divided into multiple zones, and joint traces are mapped to the corresponding zones. The joint traces are used as nodes and the relationships between traces are used as edges to construct the joint trace network for each zone. To transform complex joint distributions into a quantifiable graph structure, the process of constructing monitoring zones and joint trace networks is as follows: S31: Divide the monitoring area into six zones: the arch zone, left shoulder zone, right shoulder zone, left side zone, right side zone, and floor zone; for example... Figure 4 As shown, Figure 4 A schematic diagram of the roadway monitoring zones of this invention is shown.

[0047] More preferably, the arched area is the central area at the top of the roadway cross-section, the left and right shoulder areas are the transition areas between the arched area and the two sidewalls, the left and right sidewall areas are the sidewall areas on both sides of the roadway, and the floor area is the bottom area of ​​the roadway. Preferably, the boundaries of the monitoring zones can be adaptively adjusted according to the geometric shape of the roadway cross-section, the support arrangement, and the characteristics of the surrounding rock structure to improve the pertinence of local risk identification and support effect evaluation.

[0048] S32: Trace mapping: Based on the spatial location, endpoint coordinates and area range of the joint trace, the joint trace is mapped to the corresponding monitoring zone; when the joint trace is entirely within a single monitoring zone, it is directly assigned to the corresponding monitoring zone; when the joint trace spans two or more monitoring zones, the monitoring zone to which it belongs is determined according to the area where its main body is located or the area with the largest length proportion. Preferably, when the length proportion difference of a joint trace in multiple monitoring zones is less than a preset threshold, it can be marked as a cross-zone joint trace and participate in risk analysis simultaneously in adjacent monitoring zones. Preferably, the preset threshold for the length proportion difference is 5% to 15%, more preferably 10%.

[0049] S33: Construct a network; using the joint traces in each monitoring zone as network nodes, establish connection edges between nodes based on the intersection relationship, endpoint proximity relationship, or extension reachability relationship of the traces, forming a joint trace network for the corresponding monitoring zone, providing a foundation for subsequent analysis of joint network connectivity and evolution characteristics.

[0050] Further preferably, when two joint traces have a geometric intersection within the same monitoring zone, it is determined that there is a direct connecting edge between the two nodes; when the distance between the endpoints of the two joint traces is less than a fourth preset threshold A, it is determined that there is an endpoint-approaching connecting edge between the two nodes; preferably, the fourth preset threshold A is 5mm to 30mm, more preferably 15mm. When two joint traces can intersect or approach each other within a preset tolerance range after extending along their respective main directions, it is determined that there is an extended reachable connecting edge between the two nodes.

[0051] More preferably, the joint trace network can be represented as an undirected graph, where nodes represent joint trace objects and edges represent the connectivity relationships between joint traces. Preferably, different connection weights can be assigned to directly intersecting edges, endpoint-closed edges, and extended reachable edges to characterize the influence of different connectivity relationships on the evolution of the joint network. Preferably, the weight of directly intersecting edges is 1.0, the weight of endpoint-closed edges is 0.8, and the weight of extended reachable edges is 0.6. Through the above method, the spatial connectivity relationships between joint traces can be converted into a graph-based network structure, providing a foundation for subsequent analysis of the connectivity and evolution characteristics of the joint network.

[0052] As a preferred implementation, the monitoring partitioning and network construction parameters in S3 are shown in Table 3: Table 3: Partitioned Network Construction Parameter Table In this technical solution, firstly, the roadway monitoring area is divided into six zones: the arch zone, the left shoulder zone, the right shoulder zone, the left side zone, the right side zone, and the floor zone. This division method highly matches the roadway's stress and deformation characteristics and risk distribution patterns, making the subsequent risk assessment results more engineering-oriented and practically instructive. Secondly, in the trace mapping stage, joint traces are assigned to corresponding zones based on their spatial location and endpoint coordinates. Traces spanning multiple zones are assigned based on the principle of having the largest proportion of the main area or length. The processing rules are clear and highly operable, avoiding duplicate calculations or omissions caused by unclear trace assignments. Finally, in the network construction stage, joint traces within each zone are used as network nodes. Connection edges between nodes are established based on trace intersection relationships, endpoint proximity relationships, and extension reachability relationships. This transforms the complex spatial distribution relationship of joints into a clear and quantifiable graph theory network structure, laying a solid mathematical model foundation for subsequent quantitative analysis based on graph theory, such as the scale of connected components, node degree, network connectivity, and penetration rate.

[0053] S4: Joint network connectivity analysis and surrounding rock risk level assessment; Based on graph theory analysis of the connectivity and evolution characteristics of joint networks in each monitoring zone, dangerous joint types are identified, and the risk level of surrounding rock is classified according to multi-index thresholds (five levels from safe to high risk), with support for adaptive threshold adjustment. To achieve a five-level quantitative assessment of surrounding rock risk, the process of joint network connectivity analysis and surrounding rock risk level assessment is as follows: S41: Calculate network connectivity characteristics: Calculate the scale of connected components, node degree, network connectivity, percentage of the largest connected subgraph, key connecting paths and connectivity rate of each monitoring zone to characterize the connectivity and evolution characteristics of the joint network. Among them, network connectivity is calculated according to formula (3). The greater the network connectivity, the stronger the connectivity between joint traces within the partition, and the higher the risk of surrounding rock instability.

[0054] (3); In the formula, The number of nodes in the partitioned joint trace network. This represents the number of network edges. The penetration rate is calculated according to formula (4). The higher the penetration rate, the higher the possibility of joint clusters forming key penetration channels within the zone, and the greater the risk of local instability of the surrounding rock or block cutting instability. (4); In the formula, To determine the number of joint clusters required to form a critical through path, This represents the total number of joint clusters in the partition; As a preferred embodiment, the key connectivity path is a joint connectivity path that connects multiple dangerous joint nodes within a partition and traverses the dangerous area; the maximum connected subgraph proportion is the ratio of the number of nodes in the maximum connected subgraph to the total number of nodes in the partition; the node degree is the number of edges connecting a single joint trace to other joint traces; and the connectivity component size is the number of joint trace nodes contained in a single connected subgraph. These indicators are collectively used to characterize the overall development degree, local clustering degree, and connectivity trend of the joint network.

[0055] S42: Identify hazardous joint types; based on opening characteristics, length change, length growth rate, opening degree change, opening degree growth rate, connectivity characteristics, and location characteristics, identify hazardous joint types; the hazardous joint types include opening type, extension type, connectivity type, through type, activation type, and hazardous joints in critical locations; As a preferred option, open-type dangerous joints are joint traces with large open characteristic parameters and continuously increasing openness growth rate; extended-type dangerous joints are joint traces with continuously increasing length and continuously increasing length growth rate; connected-type dangerous joints are joint traces with increased nodality and expanded connected component scale; penetrating-type dangerous joints are joint traces that participate in the formation of key penetrating paths or whose penetration rate in the joint cluster exceeds a threshold; activated-type dangerous joints are joint traces that continue to develop or redevelop after support; critical-location dangerous joints are joint traces located in sensitive locations in the crown area, shoulder area, or sidewall and have a significant adverse impact on the stability of the surrounding rock.

[0056] S43: Classify the risk level of the surrounding rock; classify the risk level of the surrounding rock in each monitoring zone according to the geometric and evolutionary characteristic parameters of joint traces and the connectivity and evolution characteristics of joint networks; the risk level of the surrounding rock includes five levels: safe, relatively safe, relatively dangerous, dangerous, and high dangerous.

[0057] In order to achieve dynamic optimization of the assessment criteria and flexible management that is controllable by humans, the risk level of the surrounding rock is comprehensively determined based on the threshold range of the opening growth rate, length growth rate, network connectivity and penetration rate, and the threshold range of each indicator can be divided according to Table 4. Table 4: Threshold Ranges for Surrounding Rock Risk Levels Specifically as follows: Security: The growth rate of the opening is less than 5%, the growth rate of the length is less than 5%, the network connectivity is less than 0.20, and the penetration rate is less than 0.10. Relatively safe: the opening growth rate is 5% to 10%, the length growth rate is 5% to 10%, the network connectivity is 0.20 to 0.35, and the penetration rate is 0.10 to 0.20; More dangerous: the opening growth rate is 10% to 20%, the length growth rate is 10% to 20%, the network connectivity is 0.35 to 0.50, and the penetration rate is 0.20 to 0.35; Danger: The opening growth rate is 20%–35%, the length growth rate is 20%–35%, the network connectivity is 0.50–0.70, and the penetration rate is 0.35–0.50; High risk: Opening growth rate greater than 35%, length growth rate greater than 35%, network connectivity greater than 0.70, and penetration rate greater than 0.50.

[0058] The threshold ranges for each indicator level are initially defined using preset fixed thresholds, and then adaptively adjusted based on historical monitoring data, regional noise levels, and changes in support response. The adaptive adjustment range is limited to a preset allowable range, preferably not exceeding 20% ​​of the initial threshold. When it exceeds the preset allowable range, it requires manual authorization and confirmation to take effect. This mechanism, through a combination of multi-indicator comprehensive judgment and threshold adaptive adjustment, ensures both the initial consistency and stability of the evaluation standards and the ability to dynamically adapt to changes in different working conditions. Furthermore, the dual control of range limitation and manual authorization ensures the safety and controllability of the adjustment process.

[0059] In this technical solution, firstly, in the network connectivity characteristic calculation stage, two quantitative indicators, network connectivity degree and penetration rate, are introduced. Network connectivity degree describes the overall connection density of the network through the relationship between the number of nodes and the number of edges, while penetration rate reflects the degree of network penetration by the ratio of the number of joint clusters in key penetration paths to the total number of joint clusters. The two indicators complement each other, achieving a comprehensive quantitative characterization of the connectivity status of the joint network. Secondly, in the dangerous joint type identification stage, multiple dimensions such as opening characteristics, length change, length growth rate, opening degree change, opening degree growth rate, connectivity characteristics, and location characteristics are comprehensively considered. This allows for the identification of six types: opening type, extended type, connected type, penetrating type, activated type, and dangerous joints in key locations. The detailed and comprehensive classification provides a clear basis for risk tracing and targeted support. Finally, in the risk level classification stage, the surrounding rock risk is divided into five levels: safe, relatively safe, relatively dangerous, dangerous, and high dangerous. The granularity is reasonable and facilitates on-site hierarchical management and differentiated response, providing a clear risk benchmark for subsequent early warning triggering and support optimization suggestions.

[0060] S5: Support effectiveness assessment and early warning triggering; By comparing the degree of inhibition of joint trace evolution parameters before and after support, the support inhibition coefficient is calculated, the support effectiveness level is divided into five levels (from significantly effective to ineffective), and multi-level early warning and corresponding treatment measures are triggered based on the characteristic parameter threshold. To achieve closed-loop management of support effectiveness classification and multi-level risk response, the process of support effectiveness assessment and early warning triggering is as follows: S51: Calculate the support inhibition coefficient; compare the changes in joint trace length, length growth rate, opening change, opening growth rate, network connectivity change rate, and changes in the development state of dangerous joints before and after support, and calculate the support inhibition coefficient. ; through support inhibition coefficient Quantitatively assess the degree to which support inhibits joint trace expansion, opening, and network connectivity evolution; When the change in a certain evolution parameter before support is zero, to avoid division by zero errors and calculation anomalies, the following rules apply: If the change before support is zero and the change after support is also zero, the suppression effect of that parameter is considered completely effective, the corresponding contribution is zero (i.e., considered as no change requiring suppression), and the support suppression coefficient is 1; if the change before support is zero but the change after support is greater than zero, then parameter B has damaged the stability of the surrounding rock and is considered invalid, and the support suppression coefficient is 0. Furthermore, when the change in a certain parameter before support is close to zero (but not exactly zero), to avoid abnormally amplified calculation results due to an excessively small denominator, a preset minimum reference value can be used to replace the original denominator in the calculation, ensuring the stability and reliability of the support suppression coefficient. Preferably, the minimum reference value is 1 mm or 5% of the historical average of the corresponding parameter. Simultaneously, when dangerous joints continue to expand, open, or become more interconnected after support, even if the support suppression coefficient is at a moderate level, the corresponding support effectiveness level can be adjusted based on the development state of the dangerous joints.

[0061] The support inhibition coefficient is calculated according to formula (5). ; (5); In the formula, , , These represent the changes in opening, length, and network connectivity before support was installed. , , These represent the changes in opening, length, and network connectivity after support. , , The weights corresponding to the changes in opening, length, and connectivity after support are given, and As a preferred option, , , The initial values ​​were 0.35, 0.35, and 0.30, respectively.

[0062] S52: Classification of branch support effectiveness levels: based on support inhibition coefficient. Based on the development status of dangerous joints, the effectiveness of the support is classified into five levels: significantly effective, effective, average, poor, and ineffective. The support effectiveness level is determined based on the threshold range of the support inhibition coefficient. Specifically, the initial threshold range can be divided according to Table 5: Table 5: Threshold Ranges for Support Effectiveness Levels S53: Trigger an early warning signal; preferably, the first preset threshold B corresponds to a 10% increase in opening rate, the second preset threshold B corresponds to a 15% increase in length and a network connectivity of 0.50, the third preset threshold B corresponds to a penetration rate of 0.5, and the fourth preset threshold B corresponds to a poor or ineffective support effectiveness level. Based on relevant characteristic parameters and support effectiveness assessment results, trigger an early warning and corresponding handling measures: When the opening growth rate exceeds the first preset threshold B, a level one early warning is triggered, and measures to strengthen inspection, continuous monitoring, and key retesting of abnormal zones are output. When the crack length growth rate exceeds the corresponding length growth rate threshold in the second preset threshold B, and the network connectivity exceeds the corresponding network connectivity threshold in the second preset threshold B, a level two early warning is triggered, and measures such as special investigation of local areas, addition of monitoring points, and temporary reinforcement are output. When a critical breakthrough path is formed or the breakthrough rate exceeds the third preset threshold B, a level three early warning is triggered, and measures to strengthen support in key areas and restrict operations in related areas are output. When the dangerous joint continues to develop after support, or when the support effectiveness is lower than the fourth preset threshold B, an upgraded warning (level 4 warning) is triggered, and measures such as encrypted monitoring, enhanced on-site control, and, if necessary, the evacuation of personnel from the dangerous area are output.

[0063] To ensure both stability and flexibility in the level determination, in S52, the support effectiveness level is determined based on the level threshold range of the support inhibition coefficient. This threshold range is initially defined using a preset fixed threshold and is adaptively adjusted based on historical monitoring results. The adaptive adjustment range is limited to a preset allowable range; if it exceeds this range, it requires manual authorization before taking effect. This mechanism, through a two-tiered strategy of preset fixed thresholds and adaptive adjustment, ensures both the initial consistency and operability of the support effectiveness level determination, while dynamically optimizing the threshold based on historical monitoring results to adapt to different working conditions. Furthermore, the dual control of range limitation and manual authorization ensures the safety and controllability of the adjustment process.

[0064] In this technical solution, firstly, in the calculation of the support inhibition coefficient, a weighted comprehensive formula for the support inhibition coefficient is constructed by comparing the changes in joint opening, length, and network connectivity before and after support. This quantifies the inhibition effect of support on joint expansion, opening, and network connectivity evolution into a single numerical index. A specific processing rule is set when the change before support is zero, avoiding calculation anomalies and ensuring the robustness of the evaluation method. Secondly, in the classification of support effectiveness levels, effectiveness is divided into five levels—significantly effective, effective, average, poor, and ineffective—based on the magnitude of the support inhibition coefficient, providing fine granularity. The system is rationally designed to facilitate hierarchical management on-site. Furthermore, the threshold levels can be adaptively adjusted based on historical monitoring results and supplemented by manual authorization confirmation, balancing the stability and dynamic adaptability of the assessment standards. Finally, a four-level early warning mechanism has been established at the early warning triggering stage, corresponding to different risk scenarios such as exceeding the standard for opening growth rate, exceeding the standard for both length growth rate and connectivity, formation of critical through-path, and continued development of dangerous joints after support. Each level of early warning is equipped with clear handling measures, achieving closed-loop management of the entire process from monitoring and assessment to hierarchical response, providing clear and operable technical basis for on-site safety decisions.

[0065] S6: Output of integrated assessment and support optimization recommendations; By integrating the surrounding rock risk level, support effectiveness level, and early warning results, the system outputs hazardous location identification, zoned risk results, and targeted support optimization suggestions.

[0066] To achieve a closed-loop output from assessment to decision-making, the process of integrating assessment and support optimization recommendations is as follows: S61: Integrated assessment; A comprehensive assessment is conducted by taking into account the surrounding rock risk level, support effectiveness level, and early warning results; S62: Output support optimization recommendations; Output support optimization recommendations based on overall assessment results: The output support optimization suggestions and linkage strategies can be set according to Table 6: Table 6: Suggestions for Support Optimization and Coordination Strategies Specifically, when the surrounding rock risk level is safe or relatively safe, and the support effectiveness level is significantly effective or effective, the recommendation is to maintain the existing support and continue monitoring. When local joints continue to expand or connectivity increases but no critical through path is formed, a local reinforcement and support suggestion is output; as a preferred option, the local reinforcement and support suggestion includes implementing reinforcement and support on the dangerous joint concentration area, the connectivity enhancement area, or the local expansion area; When the surrounding rock risk level is relatively dangerous, dangerous, or high dangerous, and the support effectiveness level is average, poor, or ineffective, a recommendation to improve the support level is output; as a preferred option, the recommendation to improve the support level includes improving the support parameter level, enhancing the support strength, or expanding the coverage of support components; When a critical connection path is formed or a dangerous joint remains active after support, a recommendation to strengthen support and increase monitoring density in key areas is issued. As a preferred option, the recommendation to strengthen support and increase monitoring density in key areas includes strengthening support in areas where critical connections path are located, areas where dangerous joints remain active, and high-risk sensitive areas, and shortening the point cloud acquisition cycle and the on-site retest cycle.

[0067] Furthermore, preferably, when the system outputs a suggestion for local reinforcement support, the point cloud acquisition cycle for the corresponding monitoring area is shortened to 0.5 to 0.8 times the original cycle; when the system outputs a suggestion to increase the support level, the point cloud acquisition cycle for the corresponding monitoring area is shortened to 0.3 to 0.5 times the original cycle; when the system outputs a suggestion to strengthen support and increase monitoring density in key areas, the point cloud acquisition cycle for the corresponding monitoring area is shortened to 0.2 to 0.3 times the original cycle, and on-site manual verification prompts are generated simultaneously.

[0068] S63: Output identification and results; Based on the surrounding rock risk level, dangerous joint type, distribution of key penetration paths and support effectiveness level of each monitoring zone, identify dangerous parts of the surrounding rock in the roadway; generate a zone risk result map to visually display the risk level, dangerous joint distribution and key penetration paths of each zone; output dangerous part identification, zone risk results and corresponding support optimization suggestions.

[0069] As a preferred embodiment, the hazardous area identification includes high-risk joint concentration areas, key breakthrough paths, hazardous joints that remain active after support, and high-risk areas in sensitive areas such as the arch, shoulder, and sidewall. Preferably, the zoning risk result map is displayed in the form of a two-dimensional zoning color map, a three-dimensional roadway surface risk mapping map, or a hazardous joint annotation map to intuitively reflect the risk status, hazardous joint distribution, and support effect changes of each monitoring zone.

[0070] More preferably, a monitoring zone is marked as a hazardous area when it simultaneously meets any of the following conditions: 1. The surrounding rock risk level of the zone is hazardous or high-risk; 2. A critical breakthrough path is formed within the zone; 3. There are hazardous joints within the zone that continue to be activated after support; 4. The proportion of high-risk hazardous joints in the zone exceeds a preset proportion of the total number of hazardous joints in the zone. Preferably, the preset proportion is 30% to 50%, more preferably 40%.

[0071] As a preferred embodiment, the output results can be simultaneously sent to a visualization terminal, an early warning terminal, or a mine safety management platform to display the surrounding rock risk status, provide feedback on support effectiveness, locate hazardous areas, and assist in support decision-making. Preferably, the output results include at least the zoning risk level, hazardous joint type, hazardous location, support effectiveness level, early warning level, and support optimization suggestions.

[0072] Step six enables the quantitative identification of dangerous areas in the surrounding rock of the roadway, the visualization of risk results by zone, and the automatic generation of support optimization suggestions. This forms a closed-loop decision-making process of "risk identification - support evaluation - early warning response - optimization suggestions", improving the pertinence and timeliness of risk management of surrounding rock in underground mine roadways.

[0073] In this technical solution, firstly, in the integrated assessment stage, information from three dimensions—surrounding rock risk level, support effectiveness level, and early warning results—is comprehensively evaluated to avoid the one-sidedness of single-indicator evaluation and provide a comprehensive and reliable basis for subsequent decision-making. Secondly, in the support optimization suggestion output stage, differentiated suggestions are output for four typical scenarios: maintaining monitoring under safe conditions, local reinforcement during local expansion, increasing the support level when there is high risk and support failure, and strengthening support and increasing monitoring when a through path is formed. The suggestions are comprehensive and highly targeted, providing clear and actionable technical guidance for on-site personnel. Finally, in the output identification and results stage, dangerous areas are accurately identified to form a zonal risk result map, visually displaying the risk level, dangerous joint distribution, and key through paths of each zone. This transforms abstract risk data into intuitive graphical information, facilitating on-site personnel to quickly locate risk areas, understand the risk situation, and take corresponding measures, achieving a closed-loop process from data collection, analysis and evaluation to visual decision support.

[0074] This invention provides a method for assessing the risk of surrounding rock and the effectiveness of support in roadways based on the evolution of joint traces in temporal 3D point clouds. First, by setting fixed benchmark points in the stable area of ​​the roadway, a unified spatiotemporal benchmark for multi-period point cloud data is established, solving the key problem of inconsistent coordinates across different periods. Periodic acquisition combined with registration preprocessing achieves point cloud denoising, anomaly removal, and missing area marking, ensuring data quality and continuity. Marking missing areas and generating supplementary prompts effectively avoids misjudgments due to data gaps, laying a reliable data foundation for accurate identification of subsequent joint traces. Second, continuous identification of exposed structural surfaces on the roadway surface helps to accurately extract multi-dimensional features, including spatial location, length, orientation, opening degree, and various temporal evolution parameters. By constructing a comprehensive matching degree model, accurate positioning and tracking of the same joint trace in point clouds at different periods are achieved, solving the problem of cross-period data correlation. A delayed judgment mechanism is adopted for traces in missing areas, enhancing the robustness of the method under complex working conditions. Next, the tunnel monitoring area was scientifically divided into multiple functional zones, and joint traces were mapped to the corresponding zones according to their spatial location. A zoned joint trace network was constructed using joint traces as nodes, geometric intersections between traces, endpoint proximity, and reachability as edges. This transformed the complex joint distribution relationships into a quantifiable graph structure, providing a clear mathematical foundation for subsequent connectivity analysis. Subsequently, based on graph theory methods, key indicators such as the scale of connected components, node degree, network connectivity, and penetration rate of the joint network in each zone were analyzed, achieving a quantitative characterization of the connectivity and evolution characteristics of the joint network. It can identify various hazardous joint types, including open, extended, and connected joints, and classify the surrounding rock risk into five levels based on multi-index thresholds. Simultaneously, it supports adaptive adjustment of thresholds based on historical monitoring data, balancing the stability of the assessment criteria with the dynamic adaptability to on-site conditions. Then, by comparing the changes in evolution parameters of the same joint trace before and after support, the support inhibition coefficient is calculated to quantitatively evaluate the inhibition effect of support on joint expansion, opening, and network connectivity. Based on the degree of inhibition, the support effectiveness is divided into five levels. Furthermore, based on different threshold combinations of characteristic parameters such as opening rate increase, length rate increase, and network connectivity, multi-level early warning signals can be triggered, and corresponding graded treatment measures can be output, achieving closed-loop management from assessment to early warning. Finally, by integrating the surrounding rock risk level, support effectiveness level, and early warning results, dangerous areas of the roadway surrounding rock are accurately identified, and a zonal risk result map is generated. Based on different combinations of risk and effectiveness levels, targeted support optimization suggestions can be output, such as maintenance monitoring, local reinforcement, increasing the support level, or strengthening support in key areas, providing direct and operable technical basis for on-site decision-making.This invention can continuously identify, match across time periods and analyze the evolution of joint traces in roadway surrounding rock. It can achieve fully automated processing of the entire process, including identification of roadway surrounding rock joint evolution risks, dynamic evaluation of support effects, zonal early warning and output of support optimization suggestions. It can provide an integrated technical solution for underground mine roadway safety, from data acquisition, analysis and evaluation to decision support.

[0075] This method is simple to implement and highly intelligent. Through automatic point cloud acquisition and registration, intelligent joint trace identification and matching, graph theory network connectivity analysis, and quantitative calculation of support inhibition coefficient, it can efficiently achieve a comprehensive assessment of the risk level of the surrounding rock and the effectiveness of the support in roadways. It can also output appropriate graded early warnings and support suggestions based on different risk levels and support conditions. It has significant technological advancements and practical engineering value, avoiding the problems of strong subjectivity, low efficiency, and difficulty in quantification in traditional manual inspection methods. It is of great significance for ensuring safe production operations in underground mine roadways.

[0076] like Figure 5 As shown, the present invention also provides a roadway risk support assessment system based on temporal point cloud joint trace evolution, used to implement a roadway risk support assessment method based on temporal point cloud joint trace evolution, including: The point cloud acquisition module is used to periodically acquire three-dimensional point cloud data of the roadway surface in the roadway monitoring area, and at the same time record the spatial coordinate information of fixed reference markers to provide a reference for unified registration of point clouds in multiple periods; preferably, the point cloud acquisition module is also used to adjust the point cloud acquisition cycle according to the stability state of the surrounding rock, the support state and the early warning level of the roadway.

[0077] The unified registration module is used to complete the initial registration of point clouds from multiple periods based on fixed reference markers, and to perform fine registration based on overlapping areas of the point clouds to achieve a unified spatial coordinate reference for the point clouds from multiple periods. The module also performs preprocessing on the registered point cloud data, including point cloud denoising, outlier removal, monitoring area cropping, coordinate standardization, and missing area marking. When there are missing areas in a point cloud due to occlusion, support structure coverage, or data acquisition issues, the area is marked as a temporarily missing area, and a manual supplementary scanning prompt message is generated. The joint trace identification module is used to identify the joint traces of the surrounding rock from the point cloud of each period and extract the geometric and evolutionary feature parameters of the joint traces. The geometric and evolutionary feature parameters include the spatial location, endpoint coordinates, length, direction, and opening feature parameters of the joint traces, as well as the temporal evolution parameters such as length change, length growth rate, opening change, opening growth rate, and direction change. A cross-period matching module is used to achieve cross-period matching of the same joint trace based on spatial proximity, directional consistency, endpoint correspondence, and extension continuity, thereby establishing the temporal evolution relationship of the same joint trace. Preferably, the cross-period matching module is also used to mark joint traces that have failed to be stably identified as pending confirmation, and to perform delayed determination in conjunction with subsequent periodic data.

[0078] The partition modeling module is used to divide the roadway monitoring area into six monitoring partitions: the arch area, the left shoulder area, the right shoulder area, the left side area, the right side area, and the floor area. Based on the spatial location, endpoint coordinates, and area range of the joint traces, the module maps the joint traces to the corresponding monitoring partitions. Simultaneously, it uses the joint traces within each monitoring partition as network nodes and establishes connection edges between nodes based on the intersection relationship, endpoint proximity relationship, or extension reachability relationship of the traces, thereby establishing the joint trace network of the corresponding monitoring partition. The network connectivity analysis module analyzes the connectivity and evolution characteristics of the joint trace network in each monitoring zone. Based on the joint trace's opening characteristic parameters, length variation, length growth rate, opening variation, opening growth rate, connectivity characteristics, and location characteristics, it identifies hazardous joint types. Simultaneously, it assesses the surrounding rock risk level of each monitoring zone based on the threshold ranges for opening growth rate, length growth rate, network connectivity, and penetration rate. Preferably, the network connectivity analysis module calculates the connectivity component scale, node degree, network connectivity, maximum connected subgraph percentage, key penetration paths, and penetration rate of the joint trace network in each zone. Based on the geometric and evolutionary characteristic parameters of the joint traces and the connectivity and evolution characteristics of the joint network, it identifies opening-type hazardous joints, extended-type hazardous joints, connected-type hazardous joints, penetrating-type hazardous joints, activated-type hazardous joints, and hazardous joints in key locations.

[0079] The support effectiveness assessment module is used to quantitatively assess the degree of inhibition of the support on the expansion, opening, and network connectivity evolution of the same joint trace by comparing the changes in evolution parameters before and after support implementation. Based on the degree of inhibition and the development status of dangerous joints, the module determines the support effectiveness level. Preferably, the support effectiveness assessment module calculates the support inhibition coefficient based on the changes in opening, length, and connectivity before and after support, and classifies the support effectiveness into five levels: significantly effective, effective, average, poor, and ineffective, based on the support inhibition coefficient and the development status of dangerous joints.

[0080] The early warning and support suggestion output module is used to trigger corresponding early warning signals when relevant characteristic parameters reach preset thresholds, and output dangerous location identifiers, zoning risk results, and support optimization suggestions based on the surrounding rock risk level, support effectiveness level, and early warning results. Preferably, the support optimization suggestions include maintaining existing support and continuous monitoring, local reinforcement of support, increasing support level, and strengthening support and increasing monitoring in key areas. Further, the early warning and support suggestion output module is also used to output corresponding emergency response measures according to different early warning levels; wherein, a level 1 early warning corresponds to strengthening inspections, continuous monitoring, and key re-monitoring of abnormal zones; a level 2 early warning corresponds to conducting special investigations, adding monitoring points, and implementing local temporary reinforcement; a level 3 early warning corresponds to strengthening support in key areas and restricting operations in related areas; and an upgraded early warning corresponds to increasing monitoring, strengthening on-site control, and organizing the evacuation of personnel from dangerous areas when necessary.

[0081] This invention provides a roadway risk support assessment system based on the evolution of joint traces in time-series point clouds. It integrates point cloud acquisition, unified registration, joint trace identification, network connectivity analysis, risk assessment, support effectiveness evaluation, early warning, and support suggestion output. Through the point cloud acquisition module, a periodic acquisition strategy ensures the temporal comparability of the data. Simultaneously, the synchronous recording of marker point coordinates provides a reliable basis for subsequent high-precision registration. The unified registration module effectively eliminates systematic biases between data collected from different periods, ensuring the accuracy of subsequent joint trace comparison analysis and significantly improving data quality, laying a data foundation for reliable joint trace identification. Furthermore, the collaborative work of the point cloud acquisition module and the unified registration module effectively establishes a unified spatiotemporal benchmark for multi-period point clouds while ensuring data quality. The joint trace identification module automatically extracts the spatial location, endpoint coordinates, length, direction, opening characteristic parameters, and temporal evolution parameters such as length change and opening rate of joint traces from massive point cloud data, achieving a structured expression of joint information. By implementing a cross-period matching module, automatic association and tracking of joint traces from different periods are achieved. A delayed judgment mechanism is employed for traces in missing areas, enhancing the method's adaptability under incomplete data conditions. Simultaneously, the linkage between the joint trace identification module and the cross-period matching module enables continuous tracking and evolution feature extraction of joint traces. The partitioning modeling module scientifically matches the partitioning method with the roadway's stress characteristics and risk distribution patterns, making the risk assessment results more engineering-oriented. It also transforms complex joint distributions into quantifiable graph structures, providing a clear mathematical model for subsequent graph theory analysis. The network connectivity analysis module accurately characterizes joint network connectivity features, identifying various hazardous joint types. The surrounding rock risk level classification balances the stability of the assessment criteria with the dynamic adaptability to on-site conditions. Furthermore, the combination of the partitioning modeling module and the network connectivity analysis module transforms complex joint distributions into quantifiable graph networks, enabling accurate and efficient assessment of the surrounding rock risk level. By setting up the support effectiveness assessment module, the support effect can be quantitatively characterized through the support inhibition coefficient, transforming qualitative evaluation into a comparable numerical indicator. Furthermore, the finer granularity of the support effectiveness level facilitates subsequent on-site hierarchical management. The early warning and support recommendation output module enables dynamic risk response based on a multi-level early warning mechanism, simultaneously obtaining targeted support recommendations.

[0082] The system has a clear structure and high integration, enabling dynamic identification of roadway surrounding rock risks, evaluation of support effectiveness, zonal early warning, and support decision support. It significantly improves the level of safety management of surrounding rock in underground mine roadways and effectively avoids the problems of strong subjectivity, low efficiency, and difficulty in quantification in traditional manual inspection methods. It has significant technological advancement and engineering practical value for ensuring safe production operations in underground mine roadways and improving the level of surrounding rock safety management.

Claims

1. A method for assessing roadway risk support based on temporal point cloud joint trace evolution, characterized in that... It includes the following steps: S1: Multi-phase point cloud data acquisition and preprocessing; Fixed reference markers are set up in the stable area of ​​the roadway, and three-dimensional point cloud data is collected periodically. After registration and preprocessing, a unified spatiotemporal reference is established, and missing areas are marked and supplemented. S2: Joint trace recognition, feature extraction and cross-time matching; Identify the joint traces of surrounding rocks in point clouds of different periods, extract geometric and evolutionary feature parameters, achieve cross-period matching of the same joint traces through comprehensive matching degree, and delay the determination of traces in missing areas; S3: Monitoring zone and joint trace network construction; The tunnel monitoring area is divided into multiple zones, and joint traces are mapped to the corresponding zones. The joint traces are used as nodes and the relationships between traces are used as edges to construct the joint trace network for each zone. S4: Joint network connectivity analysis and surrounding rock risk level assessment; Based on graph theory analysis of the connectivity and evolution characteristics of joint networks in each monitoring zone, dangerous joint types are identified, surrounding rock risk levels are classified according to multi-index thresholds, and adaptive threshold adjustment is supported. S5: Support effectiveness assessment and early warning triggering; By comparing the degree of inhibition of joint trace evolution parameters before and after support, the support inhibition coefficient is calculated, the support effectiveness level is classified, and multi-level early warning and corresponding treatment measures are triggered based on the characteristic parameter threshold. S6: Output of integrated assessment and support optimization recommendations; By integrating the surrounding rock risk level, support effectiveness level, and early warning results, the system outputs hazardous location identification, zoned risk results, and targeted support optimization suggestions.

2. The method for assessing roadway risk support based on temporal point cloud joint trace evolution as described in claim 1, characterized in that, In S1, the multi-phase point cloud data acquisition and preprocessing process is as follows: S11: Set a reference point; Set a fixed reference point in a location where the surrounding rock is stable and unaffected by construction disturbance, as a unified spatial coordinate reference for multi-phase point clouds; S12: Collect point cloud data; periodically collect 3D point cloud data of the tunnel surface; S13: Multi-stage point cloud registration; Initial registration is performed first based on fixed reference markers, and then fine registration is performed based on overlapping areas of the point clouds; S14: Point cloud preprocessing; denoising, outlier removal, monitoring area cropping, coordinate standardization, and missing area marking of 3D point cloud data; S15: Handling missing regions; When a point cloud in a certain period is missing due to occlusion, coverage by support components, or data acquisition issues, it is marked as a temporarily missing area, and a prompt message for manual supplementary scanning is generated.

3. The roadway risk support assessment method based on temporal point cloud joint trace evolution according to claim 1, characterized in that, In S2, the process of joint trace identification, feature extraction, and cross-time matching is as follows: S21: Identify joint traces; Identify surrounding rock joint traces from point clouds of different phases; S22: Extract geometric and evolutionary characteristic parameters of joint traces; extract spatial location, endpoint coordinates, length, orientation, and opening characteristic parameters, as well as temporal evolution parameters such as length change, length growth rate, opening change, opening growth rate, and orientation change. The opening feature parameters are obtained by the normal distance between the point cloud boundaries on both sides of the joint trace. When they cannot be obtained stably, the apparent opening width is calculated by using local geometric concavity features and the distance between the crack edges, or an equivalent opening parameter is constructed as a substitute. Among them, the first is obtained according to formula (1). Opening characteristic parameters of joint traces ; （1）； In the formula, The distance between the two boundary point clouds extracted from the profile along the joint trace direction; The number of effective profiles used in the calculation; S23: Cross-period matching determination; setting a comprehensive matching degree threshold. Based on four dimensions—spatial proximity, directional consistency, endpoint correspondence, and extension continuity—normalized indices are calculated for each dimension, and then weighted and summed to obtain the comprehensive matching degree. ;when At that time, it was determined that the joint lines from two different periods met the cross-period matching condition; Among them, according to formula (2) Period Joint traces and Issue No. Joint trace line comprehensive matching degree ; （2）； In the formula, As a spatial proximity normalization index, As a directional consistency normalization index, As a normalized index for endpoint correspondence, To extend the continuous normalization index; , , , These are the weights of the spatial proximity normalization index, the directional consistency normalization index, the endpoint correspondence normalization index, and the extension continuity normalization index, respectively. ; S24: Delayed determination; when a joint trace that existed in the previous period is not identified in the later period and the corresponding area is marked as a temporarily missing area, the joint trace is marked as pending confirmation and a delayed determination is made in conjunction with subsequent periodic data.

4. A method for assessing roadway risk support based on temporal point cloud joint trace evolution as described in claim 1 or 2, characterized in that, In S3, the process of constructing the monitoring partition and joint trace network is as follows: S31: Divide the monitoring area into six zones: the arch zone, the left shoulder zone, the right shoulder zone, the left side zone, the right side zone, and the floor zone. S32: Trace mapping: Based on the spatial location, endpoint coordinates and area range of the joint trace, the joint trace is mapped to the corresponding monitoring zone; when the joint trace crosses two or more monitoring zones, the monitoring zone to which it belongs is determined according to the area where its main body is located or the area with the largest length proportion. S33: Construct a network; using the joint traces in each monitoring zone as network nodes, establish connection edges between nodes based on the intersection relationship, endpoint proximity relationship, or extension reachability relationship of the traces to form a joint trace network for the corresponding monitoring zone.

5. The method for assessing roadway risk support based on temporal point cloud joint trace evolution as described in claim 1, characterized in that, In S4, the process of joint network connectivity analysis and surrounding rock risk level assessment is as follows: S41: Calculate network connectivity characteristics: Calculate the scale of connected components, node degree, network connectivity, percentage of the largest connected subgraph, key connecting paths and connectivity rate of each monitoring zone to characterize the connectivity and evolution characteristics of the joint network. Among them, network connectivity is calculated according to formula (3). ; （3）； In the formula, The number of nodes in the partitioned joint trace network. This represents the number of network edges. The penetration rate is calculated according to formula (4). ; （4）； In the formula, To determine the number of joint clusters required to form a critical through path, This represents the total number of joint clusters in the partition; S42: Identify hazardous joint types; based on opening characteristics, length change, length growth rate, opening degree change, opening degree growth rate, connectivity characteristics, and location characteristics, identify hazardous joint types; the hazardous joint types include opening type, extension type, connectivity type, through type, activation type, and hazardous joints in critical locations; S43: Classify the risk level of the surrounding rock; classify the risk level of the surrounding rock in each monitoring zone according to the geometric and evolutionary characteristic parameters of joint traces and the connectivity and evolution characteristics of joint networks; the risk level of the surrounding rock includes five levels: safe, relatively safe, relatively dangerous, dangerous, and high dangerous.

6. The method for assessing roadway risk support based on temporal point cloud joint trace evolution according to claim 5, characterized in that, In S43, the risk level of the surrounding rock is determined based on the threshold range of the opening growth rate, length growth rate, network connectivity and penetration rate. The threshold range of each indicator is initially divided using a preset fixed threshold and is adaptively adjusted according to historical monitoring data, regional noise level and support response changes. The adaptive adjustment range is limited to a preset allowable range. When it exceeds the preset allowable range, it takes effect after manual authorization and confirmation.

7. The method for assessing roadway risk support based on temporal point cloud joint trace evolution according to claim 1, characterized in that, In S5, the process of support effectiveness assessment and early warning triggering is as follows: S51: Calculate the support inhibition coefficient; compare the changes in joint trace length, length growth rate, opening change, opening growth rate, network connectivity change rate, and changes in the development state of dangerous joints before and after support, and calculate the support inhibition coefficient. ; through support inhibition coefficient Quantitatively assess the degree to which support inhibits joint trace expansion, opening, and network connectivity evolution; if the change before and after support is zero, then... If the change before support is zero but the change after support is greater than zero, then ; The support inhibition coefficient is calculated according to formula (5). ; （5）； In the formula, , , These represent the changes in opening, length, and network connectivity before support was installed. , , These represent the changes in opening, length, and network connectivity after support. , , The weights corresponding to the changes in opening, length, and connectivity after support are given, and ; S52: Classification of Support Effectiveness Levels: Based on the support inhibition coefficient and the development status of dangerous joints, the support effectiveness levels are classified; the support effectiveness levels include five levels: significantly effective, effective, average, poor, and ineffective; S53: Triggering an early warning signal; triggering an early warning and corresponding handling measures based on relevant characteristic parameters and support effectiveness assessment results: When the opening growth rate exceeds the first preset threshold B, a level one early warning is triggered, and measures to strengthen inspection, continuous monitoring, and key retesting of abnormal zones are output. When the growth rate of crack length and network connectivity both exceed the second preset threshold B, a level-two early warning is triggered, and measures such as special investigation of local areas, addition of monitoring points, and temporary reinforcement are output. When a critical breakthrough path is formed or the breakthrough rate exceeds the third preset threshold B, a level three early warning is triggered, and measures to strengthen support in key areas and restrict operations in related areas are output. When the dangerous joint continues to develop after support, or when the support effectiveness is lower than the fourth preset threshold B, an upgraded warning is triggered, and measures such as encrypted monitoring, enhanced on-site control, and, if necessary, the evacuation of personnel from the dangerous area are output.

8. The method for assessing roadway risk support based on temporal point cloud joint trace evolution according to claim 7, characterized in that, In S52, the support effectiveness level is determined based on the level threshold range of the support inhibition coefficient. The level threshold range is initially divided using a preset fixed threshold and is adaptively adjusted based on historical monitoring results. The adaptive adjustment range is limited to a preset allowable range. When it exceeds the preset allowable range, it takes effect after manual authorization and confirmation.

9. The method for assessing roadway risk support based on temporal point cloud joint trace evolution according to claim 1, characterized in that, In S6, the process of integrating assessment and support optimization recommendations is as follows: S61: Integration Assessment; A comprehensive assessment is conducted based on the overall risk level of the surrounding rock, the effectiveness level of the support, and the early warning results. S62: Output support optimization recommendations; Output support optimization recommendations based on overall assessment results: When the surrounding rock risk level is safe or relatively safe, and the support effectiveness level is significantly effective or effective, the recommendation is to maintain the existing support and continue monitoring. When a local joint continues to expand or connectivity increases but no critical through path is formed, a local reinforcement and support suggestion is output. When the surrounding rock risk level is relatively dangerous, dangerous or high dangerous, and the support effectiveness level is average, poor or ineffective, a recommendation to increase the support level is issued. When a critical connection path is formed or dangerous joints remain active after support, recommendations are made to strengthen support and increase monitoring in key areas. S63: Output identification and results; Based on the surrounding rock risk level, dangerous joint type, distribution of key breakthrough paths and support effectiveness level of each monitoring zone, identify dangerous parts of the surrounding rock in the roadway; Generate a risk result map for each zone, visually displaying the risk level, distribution of dangerous joints, and key access paths for each zone; output the identification of dangerous parts, the risk results of each zone, and corresponding support optimization suggestions.

10. A roadway risk support assessment system based on temporal point cloud joint trace evolution, used to implement the roadway risk support assessment method based on temporal point cloud joint trace evolution as described in any one of claims 1 to 9, characterized in that, include: The point cloud acquisition module is used to periodically acquire three-dimensional point cloud data of the roadway surface in the roadway monitoring area, and at the same time record the spatial coordinate information of fixed reference marker points. A unified registration module is used to establish a unified spatial coordinate reference for point clouds from multiple periods and to preprocess the registered point cloud data. The joint trace identification module is used to identify the joint traces of the surrounding rock from the point cloud of each phase and extract the geometric and evolutionary feature parameters of the joint traces. A cross-period matching module is used to achieve cross-period matching of the same joint trace based on spatial proximity, directional consistency, endpoint correspondence and extension continuity. The partition modeling module is used to divide the roadway monitoring area into six monitoring partitions: the arch area, the left shoulder area, the right shoulder area, the left side area, the right side area, and the floor area, and to map joint traces to the corresponding monitoring partitions; at the same time, it is used to establish the joint trace network of the corresponding monitoring partitions. The network connectivity analysis module is used to analyze the connectivity and evolution characteristics of the joint trace network in each monitoring zone, identify dangerous joint types, and assess the surrounding rock risk level of each monitoring zone. The support effectiveness assessment module is used to quantitatively assess the degree of inhibition of the expansion, opening and network connectivity evolution of the joint trace by comparing the changes in evolution parameters of the same joint trace before and after support implementation, and to determine the support effectiveness level based on the degree of inhibition of joint evolution and the development status of dangerous joints. The early warning and support suggestion output module is used to trigger the corresponding early warning signal when the relevant characteristic parameters reach the preset threshold, and output the dangerous part identification, zoning risk results and support optimization suggestions according to the surrounding rock risk level, support effectiveness level and early warning results.

Images