Underground construction safety early warning method and system

Abstract

The application relates to an underground construction safety early warning method and system, and relates to the technical field of engineering construction.The application comprises the following steps: collecting geological survey data of a preset underground construction area; inputting preset engineering design parameters and the geological survey data into a preset digital twin construction module for analysis, so as to generate a real-time construction ground three-dimensional model; inputting a preset grid size and the real-time construction ground three-dimensional model into a preset risk field calculation module for analysis, so as to generate a dynamic risk field; inputting the dynamic risk field into a preset risk path prediction module for analysis, so as to generate intervention decision input data; and inputting the intervention decision input data into a preset intervention decision generation module for analysis, so as to generate a targeted decision suggestion.The application has the effect of reducing construction risks.

Description

Technical Field

[0001] This application relates to the technical field of engineering construction, and in particular to a method and system for early warning of safety during underground construction. Background Technology

[0002] The underground construction area refers to the construction area where geological surveys and construction drawings have been carried out, and the conditions for construction are met, and underground construction is about to begin or is currently underway.

[0003] In related technologies, the underground construction technology support system refers to a single-parameter alarm mechanism based on a fixed threshold used during underground construction. This system utilizes various sensors to collect data such as surrounding rock deformation and stress. When the monitored values ​​exceed a preset threshold, an alarm is triggered, providing early warning of safety accidents such as collapses. This ensures the safety of construction personnel, equipment, and project progress. The underground construction technology support system can be applied to the tunnel development stage in the early stages of mining. Because the surrounding rock in the construction area during tunnel development is often under complex geological conditions such as high ground stress, fracture zones, or faults, tunneling disturbances can easily trigger sudden instability accidents such as roof delamination, sidewall convergence, anchor bolt failure, or even roof and sidewall collapses. Therefore, using this underground construction technology support system can issue audible and visual alarms immediately, reminding on-site personnel to evacuate in time or take temporary support measures, thereby effectively reducing the accident rate.

[0004] Regarding the aforementioned technologies, on the one hand, existing system risk assessment methods are relatively simplistic, relying mainly on historical statistical data and expert experience. They lack the ability to dynamically simulate and predict the risk evolution process, resulting in early warnings during construction often focusing on post-event alerts rather than pre-event warnings. Consequently, they cannot effectively predict the dynamic propagation path and development trend of risks. On the other hand, traditional point-based monitoring and local assessment outputs mostly discrete data, making it difficult to comprehensively grasp the overall risk status of underground construction systems. This lack of systematic understanding of the spatial distribution characteristics and spatiotemporal evolution patterns of risks leads to reduced accuracy in construction risk early warnings and increased construction risks, indicating room for improvement. Summary of the Invention

[0005] To reduce construction risks, this application provides an underground construction safety early warning method and system.

[0006] Firstly, this application provides a method for early warning of safety during underground construction, employing the following technical solution: A method for early warning of safety during underground construction includes: Collect geological survey data for the pre-designated underground construction area; The preset engineering design parameters and geological survey data are input into the preset digital twin construction module for analysis to generate a real-time three-dimensional model of the construction site; Input the preset grid size and real-time 3D model of the construction site into the preset risk field calculation module for analysis to generate a dynamic risk field; The dynamic risk field is input into the preset risk path prediction module for analysis to generate intervention decision input data; The intervention decision input data is fed into a preset intervention decision generation module for analysis to generate targeted decision recommendations.

[0007] Optionally, the digital twin construction module includes the Hooke-Brown criterion and Bayesian filtering assimilation algorithm. The steps of inputting preset engineering design parameters and geological survey data into the preset digital twin construction module for analysis to generate a real-time three-dimensional model of the construction site include: Based on the Hawke-Brown criterion, engineering design parameters and geological survey data are analyzed to generate initial material constitutive parameters and an initial three-dimensional model of the construction site. Collect on-site monitoring data; The initial material constitutive parameters, the initial 3D model of the construction site, and the on-site monitoring data are input into the Bayesian filtering assimilation algorithm to update the model and generate a real-time 3D model of the construction site.

[0008] Optionally, the risk field calculation module includes a finite element discretization method, a formula for calculating the movement velocity of mesh nodes, and partial differential equations. The steps of inputting a preset mesh size and a real-time 3D model of the construction site into the preset risk field calculation module for analysis to generate a dynamic risk field include: The 3D model of the real-time construction site is divided into mesh elements based on the finite element discretization method and mesh size. Collect the current node coordinates and node movement velocity of the grid cell; Substitute the current node coordinates and node velocity into the formula for calculating the mesh node velocity to generate the node coordinates for the next time step. Collect geomechanical indicators, construction activity intensity, environmental impact factors, and population density; Based on the node coordinates and grid cells of the next time step, the geomechanical indicators, construction activity intensity, environmental impact factors and personnel distribution density are substituted into the partial differential equations for calculation to generate the risk intensity scalar of each cell; Calculate the first derivative of the risk intensity scalar for each unit to generate the risk gradient for each unit; The risk intensity scalar and risk gradient of each unit are summarized to generate a dynamic risk field.

[0009] Optionally, the risk path prediction module includes a Monte Carlo path search algorithm, a risk field intensity integral calculation formula, and a graph theory analysis algorithm. The step of inputting the dynamic risk field into the preset risk path prediction module for analysis to generate intervention decision input data includes: Find the grid risk intensity scalar in the dynamic risk field; The risk propagation path is randomly generated using the Monte Carlo path search algorithm. The grid risk intensity scalar is then substituted into the risk field intensity integral calculation formula based on the risk propagation path to generate the path occurrence probability correspondence and path occurrence probability. The probability of path occurrence is filtered based on the correspondence between the probability of path occurrence and the preset probability threshold to generate paths with significant risks. Significant risk paths are input into a graph theory analysis algorithm for analysis to generate input data for intervention decisions.

[0010] Optionally, the intervention decision generation module includes a standardized contingency plan library and a cost function. The steps of inputting intervention decision input data into the preset intervention decision generation module for analysis to generate targeted decision recommendations include: Identify the location of risk convergence points and the grid risk intensity scalar from the intervention decision input data; The intervention decision input data is entered into a standardized contingency plan library for semantic similarity calculation to generate candidate intervention strategies; The risk intensity of the risk convergence point is found in the grid risk intensity scalar based on the location of the risk convergence point. Calculate the difference between the risk intensity at the risk convergence point and the pre-set expected risk intensity after intervention to generate the expected risk reduction value; The expected risk reduction is substituted into the cost function for calculation to generate the minimum cost value; Based on the minimum cost, targeted decision-making recommendations are identified from candidate intervention strategies.

[0011] Secondly, this application provides an underground construction safety early warning system, which adopts the following technical solution: An underground construction safety early warning system includes: The data acquisition module is used to collect geological exploration data; A memory for storing a program for an underground construction safety early warning method as described in any of the above-mentioned items; The processor and the program in the memory can be loaded and executed by the processor to implement an underground construction safety early warning method as described in any of the above.

[0012] In summary, this application includes at least one of the following beneficial technical effects: 1. By inputting engineering design parameters and geological survey data into the digital twin construction module for analysis, a real-time 3D model of the construction site is obtained. The grid size and the real-time 3D model of the construction site are input into the risk field calculation module for analysis to obtain a dynamic risk field. The dynamic risk field is input into the risk path prediction module for analysis to obtain intervention decision input data. The intervention decision input data is input into the intervention decision generation module for analysis to obtain targeted decision suggestions. As a result, when underground construction is carried out based on the targeted decision suggestions, the accuracy and foresight of construction risk early warning are improved, thereby improving the safety of the construction process. 2. By finding the grid risk intensity scalar in the dynamic risk field, a risk propagation path is randomly generated using the Monte Carlo path search algorithm. The grid risk intensity scalar is then substituted into the risk field intensity integral calculation formula based on the risk propagation path to obtain the path occurrence probability correspondence and the path occurrence probability. The path occurrence probability is then filtered based on the path occurrence probability correspondence and the occurrence probability threshold to obtain significant risk paths. These significant risk paths are then input into a graph theory analysis algorithm for analysis to obtain intervention decision input data, which in turn provides data support for determining targeted decision recommendations through the subsequent intervention decision generation module. 3. By identifying the location of risk convergence points and the grid risk intensity scalar in the intervention decision input data, the intervention decision input data is input into a standardized contingency plan library for semantic similarity calculation to obtain candidate intervention strategies. Based on the location of the risk convergence points, the risk intensity of the risk convergence points is found in the grid risk intensity scalar. The difference between the risk intensity of the risk convergence points and the expected risk intensity after intervention is calculated to obtain the expected risk reduction value. Then, the expected risk reduction value is substituted into the cost function to calculate the minimum cost value. Based on the minimum cost value, targeted decision suggestions are found in the candidate intervention strategies. This improves the accuracy and foresight of construction risk warning when underground construction is carried out based on targeted decision suggestions, thereby improving the safety of the construction process. Attached Figure Description

[0013] Figure 1 This is a flowchart of an underground construction safety early warning method according to an embodiment of this application.

[0014] Figure 2 This is a flowchart of the steps in this application embodiment to input preset engineering design parameters and geological survey data into a preset digital twin construction module for analysis in order to generate a real-time three-dimensional model of the construction site.

[0015] Figure 3 This is a flowchart of the steps in this embodiment of the application to input a preset grid size and a real-time 3D model of the construction site into a preset risk field calculation module for analysis in order to generate a dynamic risk field.

[0016] Figure 4This is a flowchart of the steps in this application embodiment to input a dynamic risk field into a preset risk path prediction module for analysis in order to generate intervention decision input data.

[0017] Figure 5 This is a flowchart illustrating the steps in this application embodiment of inputting intervention decision data into a preset intervention decision generation module for analysis to generate targeted decision recommendations. Detailed Implementation

[0018] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1 to 5 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0019] This application discloses an underground construction safety early warning method. This method primarily addresses the issue of improving the accuracy and foresight of construction risk early warning. Specifically, it discloses an underground construction area, underground construction monitoring equipment, and a processing terminal. The processing terminal and the underground construction monitoring equipment are connected for data interaction. After the underground construction monitoring equipment transmits the collected geological survey data to the processing terminal, the processing terminal inputs engineering design parameters and geological survey data into a digital twin construction module for analysis to obtain a real-time 3D model of the construction site. The grid size and the real-time 3D model of the construction site are then input into a risk field calculation module for analysis to obtain a dynamic risk field. The dynamic risk field is then input into a risk path prediction module for analysis to obtain intervention decision input data. This intervention decision input data is then input into an intervention decision generation module for analysis to obtain targeted decision suggestions. The aim is to quickly and reasonably obtain targeted decision suggestions, thereby improving the accuracy and foresight of construction risk early warning when underground construction is carried out based on these targeted decision suggestions, thus enhancing the safety of the construction process.

[0020] Reference Figure 1 This application discloses an underground construction safety early warning method, which includes the following steps: Step S100: Collect geological survey data of the preset underground construction area.

[0021] Among them, geological exploration data refers to a set of data reflecting the physical and mechanical properties of rock mass and geological structure when the underground construction area traverses strata. It mainly includes the uniaxial compressive strength of rock, rock mass quality index, joint condition parameters and groundwater influence parameters. In one embodiment, the operator conducts a survey of the geological environment of the underground construction area to obtain the uniaxial compressive strength of rock, rock mass quality index, joint condition parameters and groundwater influence parameters, and inputs the survey data into the processing terminal for aggregation to obtain the geological exploration data.

[0022] The uniaxial compressive strength of rock refers to the pressure that a rock can withstand under uniform axial pressure until it is destroyed. In one embodiment, during the early geological survey of underground construction, operators drill cores to obtain multiple rock samples from the underground construction area. Each rock sample is then subjected to uniform and continuous pressure using a pressure testing machine until it is completely destroyed. The pressure applied by the pressure testing machine at the point of destruction is recorded. The uniaxial compressive strength of each rock sample is obtained by dividing all the ultimate pressures of each rock sample by the cross-sectional area of ​​the corresponding rock sample. Finally, the arithmetic mean of the uniaxial compressive strengths of all rock samples is calculated to obtain the uniaxial compressive strength of the rock.

[0023] Rock mass quality index refers to data used to quantify the integrity and degree of fragmentation of rock mass. In one embodiment, after core drilling, the operator sorts each rock sample according to the drilling depth and marks the total length of each run, the start and end depth of each run, and the length of each rock sample. Among all the rock samples, those with a length of not less than 10 cm and that are intact and undamaged are selected. The length of the selected rock sample is divided by the corresponding total length of the run to obtain the quality index of each rock mass. Then, the arithmetic mean of the quality indices of all the selected rock samples is calculated to obtain the rock mass quality index.

[0024] Groundwater impact parameters refer to parameters used to quantify the degree of influence of groundwater on the stability of surrounding rock. In one embodiment, after core drilling, the operator places two expandable rubber plugs at each borehole and injects water at a certain pressure into the rubber plugs. The water injection volume per unit time at each pressure level is measured by a flow meter. The water injection volume is divided by the corresponding pressure and then by the borehole depth to obtain the groundwater impact parameters.

[0025] Joint condition parameters refer to parameters used to describe the geometric characteristics and mechanical properties of natural cracks in rock masses. Based on joint condition parameters, the integrity of the rock mass and the ease with which surrounding rock instability or collapse propagates within the rock mass can be assessed. In one embodiment, after construction begins, operators use a 3D laser scanner to scan the tunnel face and tunnel walls to obtain point cloud data. Then, a clustering algorithm is used to automatically identify joint surfaces based on the point cloud data and calculate the number of joints, the distance between two adjacent joints, the dip angle of the joint surface to the horizontal plane, the projection direction of the joint surface normal on the horizontal plane, and the joint roughness. Finally, all the calculated data are summarized to obtain the joint condition parameters.

[0026] The underground construction area refers to the construction scope where geological surveys and construction drawings have been carried out, and underground construction is about to begin or is currently underway. In one embodiment, the underground construction scenario is the tunnel development stage before coal mining. In the early stage of coal mining, the surrounding rock is often under complex geological conditions such as high ground stress, fracture zones or faults. Tunneling disturbances can easily cause sudden instability accidents such as roof delamination, sidewall convergence, anchor bolt failure, or even roof and sidewall collapse, which can lead to casualties, equipment damage, or project delays.

[0027] Step S101: Input the preset engineering design parameters and geological survey data into the preset digital twin construction module for analysis to generate a real-time three-dimensional model of the construction site.

[0028] The real-time construction site 3D model refers to a 3D digital model that is synchronized in real time with the actual construction progress and geological survey data of the underground construction area. It is used to accurately reproduce the actual physical state of the underground construction area. By inputting engineering design parameters and geological survey data into the digital twin construction module through a processing terminal for analysis, the real-time construction site 3D model can be obtained. Specific methods are detailed in [reference needed]. Figure 2 This process provides accurate data support for subsequent risk warning and construction path determination.

[0029] Engineering design parameters refer to the set of basic parameters of the completed underground project, which are planned and designed in advance by the operators based on the site survey. They mainly include the length of the underground construction area, the burial depth of the construction site, the cross-sectional shape of the construction site, and the support type. In one embodiment, the operators input the length of the underground construction area, the burial depth of the construction site, the cross-sectional shape of the construction site, and the support type into the processing terminal according to the determined construction design drawings.

[0030] The length of the underground construction area refers to the total length of the underground construction area designed by the operators; the burial depth of the construction site refers to the vertical distance between the ground surface and the top of the underground construction area. In one embodiment, when the underground construction area is a roadway in a coal mine, the burial depth of the construction site is the vertical distance between the ground surface and the top of the roadway; the cross-sectional shape of the construction site refers to the shape of the cross-section of the underground construction area. In one embodiment, the cross-sectional shape of the construction site includes circular, horseshoe, rectangular, and elliptical shapes; the support type refers to the type of support device used to support the underground construction area and prevent it from collapsing or deforming during construction. In one embodiment, the support types include initial support type, secondary lining type, anchor spraying support type, and steel support type.

[0031] The digital twin building module refers to a module used to generate a three-dimensional digital twin model of an underground construction area based on engineering design parameters and geological survey data, and to receive multi-source monitoring data from the construction site in real time. The digital twin model dynamically evolves to simulate the physical state and behavior of the actual construction area.

[0032] Step S102: Input the preset grid size and the real-time construction site 3D model into the preset risk field calculation module for analysis to generate a dynamic risk field.

[0033] The dynamic risk field refers to the vector field of each grid cell after dividing the real-time 3D model of the construction site into multiple grid cells. The dynamic risk field transforms risk monitoring of underground construction areas from point monitoring to continuous monitoring, thus intuitively reflecting the risk evolution trend. The dynamic risk field is obtained by inputting the preset grid size and the real-time 3D model of the construction site into a preset risk field calculation module through a processing terminal for analysis. Specific methods are described in [reference needed]. Figure 3 This process provides accurate data support for subsequent risk warning and construction path determination.

[0034] The risk field calculation module is a module that connects to the digital twin construction module and is used to quantify the risk distribution in the spatial mesh of the generated real-time construction site 3D model.

[0035] The grid size refers to the side length of each grid cell when the real-time construction site 3D model is discretized into multiple grid cells. In one embodiment, the grid size is 0.5m-0.2m.

[0036] Step S103: Input the dynamic risk field into the preset risk path prediction module for analysis to generate intervention decision input data.

[0037] The intervention decision input data refers to the dataset used to support the subsequent intervention decision generation module in its decision analysis. This data is obtained by inputting the dynamic risk field into the risk path prediction module via a processing terminal. Specific methods are detailed in [reference needed]. Figure 4 This process provides data support for determining the final construction plan.

[0038] The intervention decision generation module is connected to the risk path prediction module and is used to generate and output targeted decision recommendations, including intervention timing, intervention location, and intervention measure type, based on the intervention decision input data.

[0039] The risk path prediction module is connected to the risk field calculation module and is used to predict the risk evolution path from the current construction state to potential risk events based on the dynamic risk field and through path search algorithms.

[0040] Step S104: Input the intervention decision input data into the preset intervention decision generation module for analysis to generate targeted decision recommendations.

[0041] Targeted decision-making recommendations refer to recommendations that include the timing, location, and type of intervention. These recommendations are generated by inputting the intervention decision-making data into the intervention decision generation module via a processing terminal. Specific methods are detailed in [reference needed]. Figure 5 These steps improve the accuracy and foresight of construction risk warnings and enhance safety when construction is carried out based on targeted decision-making recommendations.

[0042] The intervention decision generation module in this step is the same as the intervention decision generation module in step S103 above, and will not be described again here.

[0043] Reference Figure 2 The steps for generating a real-time 3D model of the construction site by inputting preset engineering design parameters and geological survey data into a preset digital twin construction module for analysis include: Step S200: Analyze the engineering design parameters and geological survey data based on the Hawke-Brown criterion to generate initial material constitutive parameters and an initial three-dimensional model of the construction site.

[0044] The initial material constitutive parameters refer to the set of nonlinear failure characteristic parameters of the rock mass in the underground construction area under complex stress states, thus providing data support for the subsequent determination of the real-time three-dimensional model of the construction site. The uniaxial compressive strength of the rock is found in the geological exploration data through the processing terminal, and the uniaxial compressive strength of the rock is substituted into the formula corresponding to the Hawke-Brown criterion. Calculations were performed to obtain the maximum principal stress, where, This refers to the maximum principal stress. This refers to the minimum principal stress. This refers to the uniaxial compressive strength of rock. , , This refers to empirical constants for rock masses. The internal friction angle and cohesion are then obtained by inverting the maximum principal stress and the corresponding minimum principal stress. Summarizing the internal friction angle and cohesion yields the initial material constitutive parameters. In one embodiment, the obtained maximum principal stress and corresponding minimum principal stress are fitted to a straight line using the Mohr-Coulomb strength envelope. The slope of this straight line is the internal friction angle. The intercept on the axis is the cohesive force.

[0045] The rock mass empirical constant is used to quantify the strength of rock mass under different structural characteristics. The rock mass empirical constant is the initialization parameter of the entire system. In one embodiment, the operator inputs the surrounding rock classification into the processing terminal. The processing terminal selects from the parameter database of various typical surrounding rock grades built into the system according to the input surrounding rock classification to obtain the rock mass empirical constant.

[0046] The initial three-dimensional model of the construction site refers to the benchmark static model of the underground construction area established based on engineering design parameters and geological survey data. In one embodiment, the engineering design parameters and geological survey data are input into the three-dimensional model construction algorithm through the processing terminal for data parsing and fusion. Then, the parsed and fused data is analyzed using parametric lofting algorithm, stratigraphic space interpolation algorithm and coordinate unified registration algorithm to obtain the initial three-dimensional model of the construction site.

[0047] Step S201: Collect on-site monitoring data.

[0048] Among them, the field monitoring data refers to the data set detected by sensors deployed in the underground construction area, including surrounding rock stress, convergence deformation and support structure stress. The field monitoring data can be obtained by summarizing the real-time monitored surrounding rock stress, convergence deformation and support structure stress through the processing terminal.

[0049] Surrounding rock stress refers to the direction in which the rock mass bears the maximum compressive stress. In one embodiment, the operator embeds a stress gauge in the surrounding rock borehole, reads the vibrating wire frequency of the vibrating wire stress gauge using a frequency reading instrument, and then substitutes the measured vibrating wire frequency into the factory calibration formula. By performing calculations, the surrounding rock stress can be obtained, where, This refers to the stress in the surrounding rock. This refers to the measured frequency of the vibrating string. This refers to the initial frequency of the stress gauge, in... For example, 2100Hz.

[0050] Convergence deformation refers to the distance by which the surrounding rock shrinks into the tunnel after underground construction begins. Convergence deformation is used to reflect the stability of the surrounding rock in the underground construction area. In one embodiment, the operator sets up convergence gauges at the arch, waist and sidewall positions of the cross section of the underground construction area. The convergence deformation can be obtained by measuring the relative displacement transformation between the convergence gauges.

[0051] The stress on the support structure refers to the pressure borne by the support device made of materials such as concrete or anchor rods in the underground construction area. It can reflect whether the support device has failed. In one embodiment, the operator embeds a rebar gauge in the support device and reads the reading on the rebar gauge through a processing terminal to obtain the stress on the support structure.

[0052] Step S202: Input the initial material constitutive parameters, the initial 3D model of the construction site, and the on-site monitoring data into the Bayesian filtering assimilation algorithm to update the model and generate a real-time 3D model of the construction site.

[0053] In this step, the real-time construction site 3D model is the same as the real-time construction site 3D model in step S101 above. The initial material constitutive parameters, the initial construction site 3D model and the on-site monitoring data are input into the Bayesian filtering assimilation algorithm through the processing terminal to update the model, and the real-time construction site 3D model can be obtained.

[0054] The Bayesian filtering assimilation algorithm refers to an algorithm in the digital twin construction module that can fuse initial material constitutive parameters and field monitoring data to correct the initial material constitutive parameters and achieve continuous updating of the model state. It mainly includes a state vector definition layer, a state update layer based on the unscented Kalman filter (UKF) method, and an output layer. The initial material constitutive parameters and the initial 3D model of the construction site are input into the state vector definition layer through a processing terminal for data integration to obtain the initial state vector. Then, the initial state vector is input into the state update layer to update the state vector. The state update formula based on the unscented Kalman filter (UKF) method is as follows: , In the formula, This refers to the predicted field monitoring data at time k, including surrounding rock stress, convergence deformation, and support structure stress. This refers to predicting the observation vector. This refers to process noise. This refers to observation noise. Finally, the updated state vector is reassigned to the initial 3D model of the construction site through the output layer to obtain the real-time 3D model of the construction site.

[0055] Process noise refers to the covariance matrix used to quantify the uncertainty of the initial three-dimensional model of the construction site. In one embodiment, the operator adjusts the value of process noise in real time and observes the simulated filtering results. When the filtering results are smooth, the corresponding adjustment value can be extracted to obtain the process noise.

[0056] Observation noise refers to the covariance matrix used to quantify the measurement errors of sensors deployed in the underground construction area and the degree of environmental interference during construction. In one embodiment, the operator finds the measurement accuracy, i.e. the standard deviation, in the technical manual of the sensor used, calculates the square of the measurement accuracy, and integrates the square values ​​calculated for all sensors into a diagonal matrix. Each diagonal element in this diagonal matrix is ​​the square of the measurement accuracy of each sensor. The integrated diagonal matrix is ​​the observation noise.

[0057] Reference Figure 3The steps for generating a dynamic risk field by inputting a preset grid size and a real-time 3D model of the construction site into a preset risk field calculation module for analysis include: Step S300: Divide the real-time construction site 3D model into mesh elements based on the finite element discretization method and mesh size.

[0058] In this context, a grid cell refers to a sub-region obtained by discretizing the real-time construction site 3D model. The processing terminal divides the real-time construction site 3D model into multiple sub-regions with the same grid size, a limited number, interconnected and non-overlapping, according to the finite element discretization method. This sub-region is then called a grid cell.

[0059] The finite element discretization method refers to a numerical calculation method used in the risk field calculation module to divide a real-time construction site 3D model into a finite number of interconnected and non-overlapping sub-regions. It mainly consists of a rule-setting layer, a mesh generation layer, and an output layer. The processing terminal inputs the eight-node hexahedral elements and the set mesh size requirements into the rule-setting layer. Then, the mesh generation layer discretizes the real-time construction site 3D model into boundary nodes and boundary line segments according to the set rules. Starting from the boundary, new nodes and elements are generated step by step inward. Each time a hexahedral element is generated, the system immediately checks whether the geometric quality of the element exceeds the range. If it does, the element is rejected, the node position is adjusted, and it is regenerated. The above actions are repeated until the entire computational domain is filled with elements. Finally, the output layer outputs all the generated elements to obtain the mesh element.

[0060] Step S301: Collect the current node coordinates and node movement speed of the grid cell.

[0061] The node coordinates at the current moment refer to the position vector of each node in the three-dimensional space at the current moment. In one embodiment, the coordinates are obtained by the processing terminal directly reading them from the data storage module and summarizing all the read coordinates.

[0062] Node motion velocity refers to the vector movement of each grid cell per unit time, reflecting the motion state of the grid cell itself. In one embodiment, the processing terminal assigns different motion velocities to each grid cell based on its distance from the excavation face. When the grid cell is near the excavation face, the excavation face advance speed is determined as the node motion velocity of that network node, with the velocity direction pointing in the excavation direction. When the grid cell is far from the excavation face, the distance between the grid cell and the excavation face, along with the excavation face advance speed, are substituted into the motion velocity decay formula. The corresponding node motion velocity is obtained by calculation, where, This refers to the speed of node movement. This refers to the speed at which the excavation face advances. This refers to the distance between the nodes of the grid cell and the excavation face.

[0063] Excavation face advance speed refers to the distance the tunneling face moves forward per unit time, which is used to reflect the speed of underground construction progress. In one embodiment, the excavation face advance speed is 0.5 m / h.

[0064] Step S302: Substitute the current node coordinates and node motion velocity into the grid node motion velocity calculation formula to generate the node coordinates for the next moment.

[0065] The node coordinates at the next time step refer to the new position vector of each grid node at the end of one time step in the next time step. The current node coordinates and node motion velocity are substituted into the grid node motion velocity calculation formula by the processing terminal. By performing calculations, the coordinates of the node at the next time step can be obtained, where, This refers to the speed of node movement. This refers to the coordinates of the node at the previous time step. This refers to the node coordinates at the next time step. This refers to the time step, in order to For example, =0.1h.

[0066] Step S303: Collect geomechanical indicators, construction activity intensity, environmental impact factors, and personnel distribution density.

[0067] Geomechanical indices refer to parameters used to quantify the mechanical properties and stability of surrounding rocks, including uniaxial compressive strength, rock mass quality indices, and joint condition parameters. These parameters are searched in geological exploration data via a processing terminal, and then retrieved from the data storage unit the threshold values ​​for uniaxial compressive strength, rock mass quality indices, and joint condition parameters stored by the operator. The difference between the uniaxial compressive strength threshold and the rock mass quality index is normalized to obtain the normalized uniaxial compressive strength. Similarly, the difference between the rock mass quality index threshold and the rock mass quality index is normalized to obtain the normalized rock mass quality index. Finally, the difference between the joint condition parameter threshold and the joint condition parameter is normalized to obtain the normalized joint condition parameter. The normalized uniaxial compressive strength, normalized rock mass quality index, and normalized joint condition parameter are then added together to obtain the geomechanical indices.

[0068] The uniaxial compressive strength of the rock, rock mass quality indicators, and joint condition parameters in this step are the same as those in step S100 above, and will not be repeated here.

[0069] Construction activity intensity is used to quantify the degree of impact of current construction actions on the surrounding rock. In one embodiment, the real-time tunneling thrust, tunneling torque and real-time advance speed of the tunneling machine are read by the processing terminal, and the vibration data of each blast is collected in real time by the blasting vibration monitoring instrument. Finally, all data are normalized and then added together to obtain the construction activity intensity.

[0070] The environmental impact factor is a set of parameters used to quantify the impact of the actual environment on the stability of the surrounding rock and the safety of construction in the underground construction area. It mainly includes ambient temperature and ambient humidity. In one embodiment, the ambient humidity and ambient temperature are obtained by reading the real-time monitoring data of temperature and humidity sensors deployed in the underground construction area through the processing terminal. The ambient humidity threshold and ambient temperature threshold stored by the operator are found in the data storage unit. Then, the difference between the ambient humidity and the ambient temperature threshold is normalized to obtain the normalized ambient humidity. The difference between the ambient temperature and the ambient temperature threshold is normalized to obtain the normalized ambient temperature. Finally, the normalized ambient humidity and the normalized ambient temperature are added together to obtain the environmental impact factor.

[0071] Personnel distribution density refers to the number of operators per unit area. In one embodiment, operators entering the underground construction area wear positioning tags, and the number of personnel in each grid cell is collected in real time through a processing terminal. The personnel distribution density is then obtained by dividing the number of personnel in each grid cell by the total area of ​​all grid cells.

[0072] Step S304: Based on the node coordinates and grid cells at the next time step, substitute the geomechanical indices, construction activity intensity, environmental impact factors, and personnel distribution density into the partial differential equations for calculation to generate the risk intensity scalar for each cell.

[0073] The risk intensity scalar for each grid cell refers to the normalized risk value obtained after multidimensional comprehensive analysis of each grid cell. It is used to quantify the risk level of each grid cell. The processing terminal first updates the grid cells based on the node coordinates at the next time step. In one embodiment, the processing terminal uses the Gaussian integral method to arrange 2×2×2 Gaussian points within each grid cell, calculates the Jacobian matrix and determinant of each Gaussian point, and obtains the cell volume of each grid cell by weighted summation of each Gaussian point. The surface normal vector of each grid cell is obtained through cross product calculation, and the shape function derivative matrix is ​​calculated. Each grid cell is updated based on its cell volume, surface normal vector, and shape function derivative matrix. After updating the grid cells, the processing terminal substitutes the construction activity intensity, environmental impact factor, and personnel distribution density into the partial differential equation based on the updated grid cells. By performing calculations, the risk intensity scalar for each unit can be obtained, where, This refers to a scalar value representing the risk intensity of a grid cell, ranging from [0,1]. A higher value indicates a higher risk for that grid cell. This refers to the risk diffusion coefficient tensor, used to reflect the propagation characteristics of risk in rock masses. =1.5m 2 For example, / h This refers to risk source items. , , , As the weighting coefficient, =0.6、 , For example, This refers to the intensity of construction activities. This refers to environmental impact factors. This refers to population density.

[0074] Step S305: Calculate the first derivative of the risk intensity scalar for each unit to generate the risk gradient for each unit.

[0075] Among them, the risk gradient of each unit refers to the data used to reflect the spread trend of risk in space and the severity of risk change. The direction of the risk gradient of each unit indicates the main trend of risk spread, and the magnitude represents the severity of risk change. The risk gradient of each unit can be obtained by calculating the first derivative of the risk intensity scalar of each unit through the processing terminal.

[0076] Step S306: Summarize the risk intensity scalar and risk gradient of each unit to generate a dynamic risk field.

[0077] The dynamic risk field in this step is the same as the dynamic risk field in step S102 above. The dynamic risk field can be obtained by summing up the risk intensity scalar of each unit and the risk gradient of each unit through the processing terminal.

[0078] Reference Figure 4 The steps for analyzing the dynamic risk field input into a preset risk path prediction module to generate intervention decision input data include: Step S400: Locate the grid risk intensity scalar in the dynamic risk field.

[0079] In this step, the grid risk intensity scalar is consistent with the risk intensity scalar of each element in step S304 above. The grid risk intensity scalar can be obtained by searching in the dynamic risk field through the processing terminal.

[0080] Step S401: Randomly generate risk propagation paths using the Monte Carlo path search algorithm, and calculate the path occurrence probability correspondence and path occurrence probability by substituting the grid risk intensity scalar into the risk field intensity integral calculation formula according to the risk propagation path.

[0081] The path occurrence probability correspondence refers to the relationship between each risk propagation path and its corresponding probability of occurrence. A risk propagation path is a spatial trajectory that may cause risk events such as surrounding rock instability, obtained by randomly sampling along the direction of the risk gradient of each cell in the dynamic risk field, starting from a high-risk grid cell. The processing terminal randomly generates risk propagation paths according to the Monte Carlo path search algorithm and substitutes the grid risk intensity scalar into the risk field intensity integral calculation formula based on the risk propagation path. The probability of each risk propagation path is calculated, and then a mapping table is formed by mapping the calculation results to the corresponding risk propagation paths to obtain the path occurrence probability correspondence. In the formula, P refers to the path occurrence probability. This refers to the risk transmission path. It refers to the grid risk intensity scalar.

[0082] The probability of a path occurring refers to the probability of each risk propagation path occurring, which is obtained by searching and summarizing the path occurrence probability correspondences in the processing terminal.

[0083] The Monte Carlo path search algorithm is a stochastic search algorithm based on stochastic process theory. It simulates the propagation process of risk in a dynamic risk field through a large number of random samples to generate multiple risk propagation paths. It mainly includes a starting point anchoring layer, a random sampling layer, a path generation layer, and a path output layer. The starting point anchoring layer sorts the risk intensity scalars of each unit and takes the grid unit with the largest value as the starting point of risk propagation. The random sampling layer performs a large number of random samples along the direction of the risk gradient of each unit in the dynamic risk field based on stochastic process theory to simulate the propagation process of risk in the dynamic risk field and obtain the initial risk propagation path. Then, the path generation layer connects the randomly sampled discrete points to form a risk propagation trajectory composed of a series of grid units. Finally, the path output layer outputs all risk propagation trajectories to obtain the risk propagation path.

[0084] Step S402: Filter the path occurrence probability according to the path occurrence probability correspondence and the preset occurrence probability threshold to generate significant risk paths.

[0085] Among them, a significant risk path refers to a risk propagation path whose probability of occurrence is greater than the probability threshold. The processing terminal filters the probability of occurrence of paths according to the probability threshold to obtain the probability of occurrence of paths exceeding the threshold. Then, based on the probability of occurrence of paths exceeding the threshold, the path occurrence probability is searched in the path occurrence probability correspondence and summarized to obtain the significant risk path.

[0086] The probability threshold is a probability benchmark value used to screen risk propagation paths. In one embodiment, the probability threshold is 0.7.

[0087] Step S403: Input the significant risk path into the graph theory analysis algorithm for analysis to generate intervention decision input data.

[0088] Among them, the intervention decision input data refers to the data set covering all dimensions of underground construction risk after risk evolution, topological analysis and feature extraction of significant risk paths, including grid risk intensity scalar, risk convergence point, risk divergence point, risk propagation bottleneck location, risk propagation critical path, risk path directed graph and visualized 3D model of construction area.

[0089] A directed graph of risk paths is a directed graph that reflects risk states and state transition relationships. By inputting significant risk paths into a graph theory analysis algorithm through a processing terminal, a directed graph of risk paths can be obtained. Among them, vertex set This indicates the risk status, including typical risk modes such as surrounding rock instability, support failure, and water inrush and mudslide. The weights represent the state transition relationships and the transition probabilities.

[0090] A risk convergence point is the coordinate of the intersection of multiple significant risk paths in a directed graph of risk paths; it is a vertex with an in-degree of at least three times its out-degree. A risk divergence point is the starting point of risk propagation in a directed graph of risk paths; multiple significant risk paths will originate from this point, and it is a vertex with an out-degree of at least three times its in-degree. A risk propagation bottleneck is a vertex that multiple paths in a significant risk path will pass through; it is a vertex with extremely high betweenness centrality in a directed graph of risk paths. A risk propagation critical path is the path with the highest probability of occurrence and the greatest degree of risk accumulation among the significant risk paths.

[0091] A visualized 3D model of a construction area refers to a 3D model rendered on a real-time 3D model of the construction site, which visualizes risk convergence points, risk divergence points, risk propagation bottlenecks, and critical paths of risk propagation.

[0092] Graph theory analysis algorithms, based on graph theory, transform significant risk paths into directed topological graphs, enabling topological analysis of risk propagation relationships. They primarily consist of a foundational construction layer, a feature extraction layer, a risk target localization layer, a risk classification and dimensionality reduction layer, and a decision output layer. The foundational construction layer transforms significant risk paths into a directed weighted graph. The feature extraction layer calculates the in-degree and out-degree of each vertex using basic degree features. It then calculates the betweenness centrality and proximity centrality of each vertex using centrality features, and calculates strong and weak connectivity components using connectivity features. The in-degree, out-degree, betweenness centrality, proximity centrality, and strong / weak connectivity components of each vertex are then aggregated to obtain a topological feature quantification dataset of the vertices and edges of the directed weighted graph. Finally, the risk target localization layer selects vertices with an in-degree at least three times their out-degree from this topological feature quantification dataset and identifies them as risks. The risk propagation bottleneck is identified by selecting vertices with an out-degree of at least three times their in-degree as convergence points. Then, the top 5% of vertices in terms of betweenness centrality are selected as risk propagation bottlenecks. Directed edge sequences are filtered based on edge weights and path frequency to obtain critical paths for risk propagation. A risk classification dimensionality reduction layer, based on graph spectral decomposition theory, clusters and stratifies significant risk paths to obtain stratified significant risk paths. Finally, a decision output layer maps the risk convergence points, risk propagation bottlenecks, and critical paths from the analysis results of the first four layers onto a real-time 3D model of the construction site, highlighting and visualizing them to obtain a visualized 3D model of the construction area. The analysis results from the first four layers are then structured and integrated to form a dataset containing risk spatial distribution, evolution patterns, core prevention and control targets, and critical paths, which is recorded as the intervention decision input data. A risk propagation topology analysis report is then output.

[0093] Reference Figure 5 The steps involved in inputting intervention decision data into a pre-defined intervention decision generation module for analysis to generate targeted decision recommendations include: Step S500: Locate the risk convergence point and the grid risk intensity scalar in the intervention decision input data.

[0094] In this step, the location of the risk convergence point is the same as that in step S402 above, and the grid risk intensity scalar in this step is the same as that in step S402 above. The location of the risk convergence point and the grid risk intensity scalar can be obtained by searching in the intervention decision input data through the processing terminal.

[0095] Step S501: Input the intervention decision input data into the standardized contingency plan library for semantic similarity calculation to generate candidate intervention strategies.

[0096] Among them, candidate intervention strategies refer to a set of standard intervention plans found from a standardized contingency plan library that are highly adaptable to the current risk scenario. The data from the intervention decision input data is integrated into a risk feature vector through a processing terminal, and the standard feature vector of each standard intervention plan in the standardized contingency plan library is extracted and calculated using a semantic similarity formula. Calculate the similarity between the risk feature vector and the standard feature vector of each standard intervention plan, and filter the standard intervention plans based on the similarity. The candidate intervention strategies are obtained by summarizing the standard intervention plans with a similarity close to 1. Where, This refers to similarity, with values ​​ranging from [0,1]. This refers to the risk feature vector. This refers to the standard feature vector of each standard intervention plan.

[0097] A standardized contingency plan library refers to a library that stores standard intervention plans that match different risk types and engineering scenarios. Each plan contains structured information such as applicable conditions, implementation steps, resource requirements, expected results, and success stories.

[0098] Step S502: Find the risk intensity of the risk convergence point in the grid risk intensity scalar based on the location of the risk convergence point.

[0099] Among them, the risk intensity of the risk convergence point refers to the grid risk intensity scalar corresponding to the location of the risk convergence point. The risk intensity of the risk convergence point can be obtained by the processing terminal searching in the grid risk intensity scalar according to the network number corresponding to the location of the risk convergence point.

[0100] Step S503: Calculate the difference between the risk intensity at the risk convergence point and the preset expected risk intensity after intervention to generate the expected risk reduction value.

[0101] The expected risk reduction value is the reduction in risk intensity at the risk convergence point after implementing risk intervention measures. This is obtained by subtracting the expected risk intensity after intervention from the risk intensity at the risk convergence point through the processing terminal.

[0102] The expected post-intervention risk intensity refers to the risk intensity that the risk convergence point is expected to reach after the implementation of risk intervention measures. In one embodiment, the expected post-intervention risk intensity is 0.3.

[0103] Step S504: Substitute the expected risk reduction value into the cost function to calculate the minimum cost value.

[0104] Here, the minimum cost value refers to the value with the smallest cost value among the candidate intervention strategies. The expected risk reduction value is substituted into the cost function through processing the terminal. By performing calculations, the minimum cost can be obtained, where, This refers to intervention costs, including the cost of materials, labor, and equipment usage. This refers to the expected reduction in risk. This refers to the impact on the construction period, used to quantify and assess the time required to implement measures and their delaying effect on subsequent processes. This refers to weighting coefficients, which sum to 1 and are dynamically adjusted based on project priority, such as assigning weights during the emergency rescue phase. Larger values ​​(0.5-0.7) are assigned during the rush work phase. Larger values ​​(0.4-0.6) are observed during the cost control phase, at which point the value is increased. The weights are calculated, and after calculating the cost value of all candidate intervention strategies, all cost values ​​are sorted, and the smallest value is the minimum cost value.

[0105] Step S505: Identify targeted decision-making recommendations from candidate intervention strategies based on minimum cost.

[0106] The targeted decision-making suggestions in this step are consistent with those in step S104 above. The targeted decision-making suggestions can be obtained by searching the candidate intervention strategies based on the minimum cost value through the processing terminal.

[0107] Based on the same inventive concept, embodiments of this application provide an underground construction safety early warning system, including: The data acquisition module is used to collect geological survey data, on-site monitoring data, current node coordinates, node movement speed, geomechanical indicators, construction activity intensity, environmental impact factors, and personnel distribution density. A memory used to store a program for an underground construction safety early warning method; The processor can load and execute programs in memory to implement an underground construction safety early warning method.

[0108] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0109] This application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a method for early warning of underground construction safety.

[0110] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

[0111] Based on the same inventive concept, this application provides a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor to provide an underground construction safety early warning method.

[0112] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0113] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for early warning of safety during underground construction, characterized in that, include: Collect geological survey data for the pre-designated underground construction area; The preset engineering design parameters and geological survey data are input into the preset digital twin construction module for analysis to generate a real-time three-dimensional model of the construction site; Input the preset grid size and real-time 3D model of the construction site into the preset risk field calculation module for analysis to generate a dynamic risk field; The dynamic risk field is input into the preset risk path prediction module for analysis to generate intervention decision input data; The intervention decision input data is fed into a preset intervention decision generation module for analysis to generate targeted decision recommendations.

2. The underground construction safety early warning method according to claim 1, characterized in that, The digital twin construction module includes the Hooke-Brown criterion and Bayesian filtering assimilation algorithm. The steps of inputting preset engineering design parameters and geological survey data into the preset digital twin construction module for analysis to generate a real-time three-dimensional model of the construction site include: Based on the Hawke-Brown criterion, engineering design parameters and geological survey data are analyzed to generate initial material constitutive parameters and an initial three-dimensional model of the construction site. Collect on-site monitoring data; The initial material constitutive parameters, the initial 3D model of the construction site, and the on-site monitoring data are input into the Bayesian filtering assimilation algorithm to update the model and generate a real-time 3D model of the construction site.

3. The underground construction safety early warning method according to claim 1, characterized in that, The risk field calculation module includes the finite element discretization method, the formula for calculating the movement velocity of mesh nodes, and partial differential equations. The steps for generating a dynamic risk field by inputting a preset mesh size and a real-time 3D model of the construction site into the preset risk field calculation module for analysis include: The 3D model of the real-time construction site is divided into mesh elements based on the finite element discretization method and mesh size. Collect the current node coordinates and node movement velocity of the grid cell; Substitute the current node coordinates and node velocity into the formula for calculating the mesh node velocity to generate the node coordinates for the next time step. Collect geomechanical indicators, construction activity intensity, environmental impact factors, and population density; Based on the node coordinates and grid cells at the next time step, geomechanical indices, construction activity intensity, environmental impact factors, and personnel distribution density are substituted into partial differential equations for calculation to generate risk intensity scalars for each cell. Calculate the first derivative of the risk intensity scalar for each unit to generate the risk gradient for each unit; The risk intensity scalar and risk gradient of each unit are summarized to generate a dynamic risk field.

4. The underground construction safety early warning method according to claim 1, characterized in that, The risk path prediction module includes a Monte Carlo path search algorithm, a risk field intensity integral calculation formula, and a graph theory analysis algorithm. The steps of inputting the dynamic risk field into the preset risk path prediction module for analysis to generate intervention decision input data include: Find the grid risk intensity scalar in the dynamic risk field; The risk propagation path is randomly generated using the Monte Carlo path search algorithm. The grid risk intensity scalar is then substituted into the risk field intensity integral calculation formula based on the risk propagation path to generate the path occurrence probability correspondence and path occurrence probability. The probability of path occurrence is filtered based on the correspondence between the probability of path occurrence and the preset probability threshold to generate paths with significant risks. Significant risk paths are input into a graph theory analysis algorithm for analysis to generate input data for intervention decisions.

5. The underground construction safety early warning method according to claim 1, characterized in that, The intervention decision generation module includes a standardized contingency plan library and a cost function. The steps of inputting intervention decision input data into the preset intervention decision generation module for analysis to generate targeted decision recommendations include: Identify the location of risk convergence points and the grid risk intensity scalar from the intervention decision input data; The intervention decision input data is entered into a standardized contingency plan library for semantic similarity calculation to generate candidate intervention strategies; The risk intensity of the risk convergence point is found in the grid risk intensity scalar based on the location of the risk convergence point. Calculate the difference between the risk intensity at the risk convergence point and the pre-set expected risk intensity after intervention to generate the expected risk reduction value; The expected risk reduction is substituted into the cost function for calculation to generate the minimum cost value; Based on the minimum cost, targeted decision-making recommendations are identified from candidate intervention strategies.

6. An underground construction safety early warning system, characterized in that, include: The data acquisition module is used to collect geological exploration data; A memory for storing a program for an underground construction safety early warning method as described in any one of claims 1 to 5; The processor and the program in the memory can be loaded and executed by the processor to implement the underground construction safety early warning method as described in any one of claims 1 to 5.

Images