Comprehensive evaluation and emergency disposal system and method for post-disaster accumulation body of underground engineering

Abstract

The invention discloses an underground engineering post-disaster accumulation body comprehensive evaluation and emergency disposal system and a method thereof.The method comprises the following steps that a geological database is pre-stored through a tunnel site edge computing terminal, multi-source data such as three-dimensional point cloud, images, water pressure and surrounding rock deformation are collected in real time, key parameters are obtained through preprocessing and convolutional neural network recognition, and the key parameters are subjected to comprehensive evaluation and emergency disposal; and a mechanism parameter set is generated by fusing formation cause information, a numerical simulation engine is driven to complete fluid-solid coupling calculation, and an equipment control instruction is automatically generated and issued after the instability risk is evaluated. According to the method, full-chain cooperation of perception, modeling, simulation, threshold triggering and execution of the post-disaster accumulation body is achieved, the systematicness and integrity of emergency disposal are greatly improved, and the overall effect of underground engineering post-disaster rescue is effectively guaranteed.

Description

Technical Field

[0001] This invention relates to the field of emergency response and intelligent monitoring technology for underground engineering disasters, specifically to a comprehensive evaluation and emergency response system and method for post-disaster accumulation of underground engineering structures. Background Technology

[0002] During tunnel construction or operation, factors such as high ground stress, fault fracture zones, water-rich strata, soft rock creep, and deterioration of rock mass structure can easily lead to disasters such as surrounding rock instability, support failure, and sudden water and mud inrushes. After these disasters, deposits may form and continue to deform, seep, or experience secondary instability. Current emergency response methods generally rely on manual reconnaissance and experience-based judgment, which presents the following technical problems: 1) On-site information is difficult to quantify objectively: the morphology, pore structure, seepage and material composition of the accumulation body mostly rely on visual inspection or local sampling. It is difficult to obtain repeatable three-dimensional geometric and physical property parameters in low visibility, high humidity and high risk environments, resulting in the omission of key features (such as local seepage concentration, weak bedding planes, plastic folds).

[0003] 2) The modeling and analysis chain is not closed: Although technologies such as 3D laser scanning, photogrammetry or numerical simulation have been introduced, the scanned data is mostly used for visualization and it is difficult to automatically generate a computable geometric model; the material identification results often remain at the information level and do not form an automatic configuration of material constitutive parameters and boundary conditions; as a result, the conversion between numerical analysis and on-site treatment still relies on manual conversion, resulting in insufficient efficiency and consistency.

[0004] 3) Delayed response to engineering incidents: The process from investigation and consultation to issuing response orders is lengthy, and the post-disaster debris may continue to seep or deform during the waiting period, increasing the probability of secondary instability and threatening the safety of personnel and equipment. Especially in tunnel environments, reliable on-site signals and automated interfaces are needed to support the access control of mechanical equipment and the triggering of personnel evacuation.

[0005] Therefore, there is an urgent need for a technical solution that is based on multi-source physical measurement, with automated numerical modeling and multi-field coupled simulation as the core, and with equipment control signal output as the endpoint, to realize an integrated closed loop of "perception-modeling-simulation-threshold triggering-execution" for post-disaster debris, thereby improving the timeliness and safety of emergency response. Summary of the Invention

[0006] In view of this, the purpose of the present invention is to provide a comprehensive evaluation and emergency response system and method for post-disaster accumulation of underground engineering structures, so as to solve the technical problems mentioned in the prior art.

[0007] A comprehensive evaluation and emergency response method for post-disaster debris in underground engineering projects, the method comprising the following steps: A geological database is generated by acquiring stratigraphic genetic information of the tunnel at the node to be monitored, and pre-stored in an edge computing terminal deployed at the tunnel site. The edge computing terminal automatically executes an emergency response procedure according to a pre-defined control program. The emergency response procedure is set as follows: Real-time acquisition of at least one of the following: three-dimensional point cloud data, on-site image data, water pressure data, and surrounding rock deformation data of the post-disaster debris body to be collected. After preprocessing at least one set of the acquired three-dimensional point cloud data, a local coordinate system is constructed, and the geometric morphological parameters of the part of the post-disaster accumulation body to be collected are calculated using the preprocessed three-dimensional point cloud data in the local coordinate system. The acquired on-site image data is input into a pre-trained convolutional neural network model. The pre-trained convolutional neural network model is configured to: obtain the probability value corresponding to each material category based on the input on-site image data, and determine the material category with the highest probability value as the material composition parameter of the part to be collected from the post-disaster deposit and output it. The geological database pre-stored in the edge computing terminal is invoked to read the stratigraphic genesis information of the tunnel corresponding to the mileage of the post-disaster deposit to be collected, and combined with the geometric morphology parameters and material composition parameters, a set of mechanism parameters characterizing the disaster formation mechanism is generated. Based on the aforementioned mechanism parameter set, a simulation configuration parameter set is automatically generated to drive the numerical simulation engine to construct a numerical model of the accumulation body, and the fluid-structure interaction values ​​of the numerical model of the accumulation body are calculated to obtain the distribution characteristics of the displacement field, stress field and seepage field of the part of the post-disaster accumulation body to be collected. Key physical quantities are extracted from the distribution characteristics of displacement field, stress field and seepage field of the post-disaster deposit to be collected and compared with preset physical threshold conditions to generate instability mode parameters and risk level parameters with correlation to characterize the instability characteristics of the deposit. Based on the instability mode parameters and the risk level parameters, control command parameters for the corresponding devices are automatically generated and sent to the corresponding devices through the communication interface to drive them to execute the corresponding emergency measures.

[0008] Optionally, the method for preprocessing the three-dimensional point cloud data specifically includes: The acquired 3D point cloud data is subjected to statistical filtering and multi-site registration processing.

[0009] Optionally, the simulation configuration parameter set includes at least geometric model parameters, material constitutive model selection parameters, and boundary load configuration parameters.

[0010] Optionally, the simulation configuration parameter set is set as the input interface parameter of the numerical simulation engine, which is used to trigger the numerical simulation engine to retrieve the corresponding material constitutive model data, boundary loads and initial conditions related to the cause of the disaster, and the construction constraint parameters of the numerical model of the accumulation body from the preset configuration file.

[0011] Optionally, the construction constraint parameters of the numerical model of the accumulation body include at least one of the following: calculation step size, convergence criterion, and safety factor.

[0012] Optionally, the geometric modeling method of the numerical model of the accumulation body is adapted and selected based on the determination results of the geometric morphology parameters, which include: Determine the partitioning method of the initial geometric model of the accumulation body, determine the mesh type or mesh density of different regions, and determine at least one of the following: the location and range of the contact surface, slip surface or weak surface in the simulation model.

[0013] Optionally, at least one of the following can be calibrated in real time based on the risk level parameter: the reduction factor of the material constitutive model selection parameter, the loading amplitude or loading rate of the boundary load configuration parameter, and the safety factor or stability criterion threshold during the simulation calculation process.

[0014] Optionally, when the measured water pressure value in the water pressure data of the post-disaster deposit to be collected and / or the measured deformation rate in the surrounding rock deformation data exceeds a preset physical threshold, a correction mechanism is automatically triggered. The correction mechanism is configured to dynamically adjust the permeability-related parameters or rheology-related parameters of the simulation configuration parameter set according to the risk level parameter.

[0015] Optionally, the control command parameters of the device include: The control signal is at least one of the following: the operating permit status of the slag removal machinery, the start / stop status of the grouting equipment, and the trigger status of the audible and visual alarm device.

[0016] A comprehensive evaluation and emergency response system for post-disaster accumulations in underground engineering projects, used to implement the methods described above, the system comprising: The data acquisition unit is configured to acquire in real time at least one of the following: three-dimensional point cloud data of the post-disaster deposit to be collected, on-site image data, water pressure data, and surrounding rock deformation data. An edge computing unit, connected to the data acquisition unit, includes a geological database, a point cloud processing module, a material identification module, a mechanism parameter generation module, a simulation configuration parameter generation module, and a multi-field coupled simulation module. The geological database is configured to contain stratigraphic genetic information of the tunnels at the monitoring nodes. The point cloud processing module is configured to preprocess at least one set of acquired 3D point cloud data to construct a local coordinate system, and use the preprocessed 3D point cloud data in the local coordinate system to calculate the geometric morphological parameters of the post-disaster deposit to be collected. The material identification module is configured to input the acquired field image data into a pre-trained convolutional neural network model. The pre-trained convolutional neural network model is configured to obtain the probability value corresponding to each material category based on the input field image data, and determine the material category with the highest probability value as the material composition parameter of the post-disaster deposit to be collected. The output is as follows: The mechanism parameter generation module is configured to: call the geological database pre-stored in the edge computing terminal, read the stratigraphic genetic information of the tunnel corresponding to the mileage of the post-disaster deposit to be collected, and combine the geometric morphology parameters and material composition parameters to generate a mechanism parameter set characterizing the disaster formation mechanism; The simulation configuration parameter generation module is configured to: automatically generate a simulation configuration parameter set based on the mechanism parameter set; The multi-field coupling simulation module is configured to: input the simulation configuration parameter set to drive the numerical simulation engine to build a numerical model of the deposit, and calculate the fluid-structure interaction values ​​of the numerical model of the deposit to obtain the distribution characteristics of the displacement field, stress field and seepage field of the post-disaster deposit to be collected; and extract key physical quantities from the distribution characteristics of the displacement field, stress field and seepage field of the post-disaster deposit to be collected and compare them with preset physical threshold conditions to generate instability mode parameters and risk level parameters with correlation to characterize the instability characteristics of the deposit; The control execution unit includes a programmable logic controller (PLC) and an actuator drive module. The PLC is configured to automatically generate control command parameters for the corresponding device based on the instability mode parameters and the risk level parameters. The actuator drive module is configured to receive the control command parameters issued by the PLC to control the corresponding device to execute corresponding emergency measures. A communication unit is used to establish a communication link between the edge computing unit and the control execution unit via an industrial Ethernet or wireless communication module.

[0017] The beneficial effects that this invention can produce include: 1. The comprehensive evaluation and emergency response system and method for post-disaster accumulation in underground engineering provided by this invention constructs a closed-loop system encompassing "real-time monitoring, data processing, mechanism analysis, numerical simulation, risk assessment, automatic response, and model calibration," ensuring seamless data connection and dynamic feedback across all stages. Through the correlation and adaptation of mechanism parameter sets and simulation configuration parameter sets, as well as the reverse calibration of the model using risk parameters, the entire evaluation and response process forms an adaptive optimization loop. This allows for continuous monitoring of the accumulation's catastrophic evolution and dynamic adjustment of the emergency response process. Compared to traditional models where monitoring and response are disconnected, and evaluation and execution are fragmented, this invention achieves full-chain collaboration of "perception-modeling-simulation-threshold triggering-execution" for post-disaster accumulation, significantly improving the systematic nature and completeness of emergency response and effectively ensuring the overall effectiveness of post-disaster rescue efforts in underground engineering.

[0018] 2. This invention enables local acquisition, processing, and analysis of multi-source monitoring data, such as 3D point clouds, on-site images, water pressure, and surrounding rock deformation data, by pre-storing the geological database on the edge computing terminal at the tunnel site. This eliminates the need for remote server transmission and computation, significantly reducing data transmission latency and computational cost. Compared to traditional offline analysis methods, the entire response time from data acquisition to evaluation and analysis to command issuance can be shortened from hours to minutes. This allows for rapid capture of the dynamic disaster characteristics of post-disaster deposits, securing crucial time for emergency response and effectively mitigating the risk of secondary disasters caused by delayed response.

[0019] 3. This invention employs a risk-driven model parameter calibration mechanism, dynamically adjusting key parameters such as the material constitutive model reduction factor and boundary load amplitude / rate based on real-time risk assessments. Simultaneously, when measured water pressure or surrounding rock deformation rates exceed preset thresholds, dynamic corrections to permeability and rheology-related parameters are automatically triggered. Compared to traditional numerical models with fixed parameters, the system's configured accumulation body numerical model can adapt in real-time to the dynamic catastrophe process of post-disaster accumulation bodies, effectively solving the problem of model-actual-condition disconnect in complex disaster environments and significantly improving the reliability and timeliness of instability prediction.

[0020] 4. This invention automatically generates control commands such as permits for cleaning machinery operation, start / stop of grouting equipment, and triggering of audible and visual alarms based on instability modes and risk level parameters. These commands are then directly sent to the corresponding equipment via a standardized communication interface, achieving automated execution of emergency measures. No manual on-site judgment or operation is required, avoiding the subjectivity and experience limitations of human decision-making. Furthermore, it allows for precise matching of treatment measures based on the risk level of different areas of the debris mass, such as differentiated grouting intensity and targeted cleaning range, avoiding over- or under-treatment, improving emergency response efficiency and resource utilization efficiency, while ensuring the safety of rescue personnel. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a logical diagram illustrating the generation of device control command parameters based on instability mode parameters and risk level parameters according to the present invention. Figure 3 This is a schematic diagram of the system architecture of the present invention; In the diagram: 1. Data acquisition unit, 2. Edge computing unit, 21. Geological database, 22. Point cloud processing module, 23. Material identification module, 24. Mechanism parameter generation module, 25. Simulation configuration parameter generation module, 26. Multi-field coupling simulation module, 3. Control execution unit, 31. Programmable logic controller, 32. Actuator drive module, 4. Communication unit. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 As shown, this invention provides a method for comprehensive evaluation and emergency response to post-disaster debris accumulation in underground engineering projects. The method includes the following steps: A geological database 21 is generated by acquiring stratigraphic genetic information of the tunnel at the node to be monitored, and pre-stored in an edge computing terminal deployed at the tunnel site. This edge computing terminal includes a data acquisition unit 1 and an edge computing unit 2. The edge computing terminal automatically executes an emergency response procedure according to a pre-defined control program. The emergency response procedure is set as follows: Real-time acquisition of at least one of the following: three-dimensional point cloud data, on-site image data, water pressure data, and surrounding rock deformation data of the post-disaster debris body to be collected; wherein, three-dimensional point cloud data and on-site image data of the post-disaster debris body to be collected are simultaneously collected by a three-dimensional laser scanner and an industrial camera, and water pressure data and surrounding rock deformation data of the corresponding location are collected by a water pressure sensor and a convergence meter. After preprocessing at least one set of three-dimensional point cloud data, a local coordinate system is constructed, and the geometric morphological parameters of the post-disaster deposit to be collected are calculated using the preprocessed three-dimensional point cloud data in the local coordinate system. The geometric morphological parameters include the porosity, aspect ratio, surface fractal dimension, number of layers, and number of steps of the post-disaster deposit to be collected. The acquired on-site image data is input into a pre-trained convolutional neural network model. The pre-trained convolutional neural network model is configured to: obtain the probability value corresponding to each material category based on the input on-site image data, and determine the material category with the highest probability value as the material composition parameter of the part to be collected in the post-disaster deposit and output it. The geological database 21 pre-stored in the edge computing terminal is called to read the stratigraphic genesis information of the tunnel corresponding to the mileage of the post-disaster deposit to be collected, and combined with geometric morphology parameters and material composition parameters to generate a set of mechanism parameters characterizing the disaster formation mechanism. The simulation configuration parameter set is automatically generated based on the mechanism parameter set to drive the numerical simulation engine to build a numerical model of the accumulation body, and the fluid-structure interaction value of the accumulation body numerical model is calculated to obtain the distribution characteristics of the displacement field, stress field and seepage field of the part of the post-disaster accumulation body to be collected, which can be used as the physical basis for the generation of subsequent equipment control commands. Key physical quantities are extracted from the distribution characteristics of displacement field, stress field and seepage field of the post-disaster deposit to be collected and compared with preset physical threshold conditions to generate instability mode parameters and risk level parameters with correlation to characterize the instability characteristics of the deposit. Based on the instability mode parameters and risk level parameters, the system automatically generates control command parameters for the corresponding equipment and sends them to the corresponding equipment through the communication interface to drive it to execute the corresponding emergency measures.

[0024] Furthermore, the method for preprocessing 3D point cloud data specifically includes: performing statistical filtering and multi-site registration on the acquired 3D point cloud data. Specifically, the statistical filtering is configured to statistically remove outliers and perform radius filtering for noise reduction in the 3D point cloud data.

[0025] Furthermore, the simulation configuration parameter set includes at least geometric model parameters, material constitutive model selection parameters, and boundary load configuration parameters.

[0026] Furthermore, the simulation configuration parameter set is set as the input interface parameter of the numerical simulation engine, which is used to trigger the numerical simulation engine to retrieve the corresponding material constitutive model data, boundary loads and initial conditions related to the cause of disasters, and construction constraint parameters of the numerical model of the accumulation body from the preset configuration file.

[0027] Furthermore, the construction constraint parameters of the numerical model of the accumulation body include at least one of the following: calculation step size, convergence criterion, and safety factor.

[0028] Furthermore, the geometric modeling method of the numerical model of the accumulation body is adapted and selected based on the determination results of the geometric morphology parameters, which include: Determine the partitioning method of the initial geometric model of the accumulation body, determine the mesh type or mesh density of different regions, and determine at least one of the following: the location and range of the contact surface, slip surface or weak surface in the simulation model.

[0029] Furthermore, based on the risk level parameter, at least one of the following is calibrated in real time: the reduction factor of the material constitutive model selection parameter, the loading amplitude or loading rate of the boundary load configuration parameter, and the safety factor or stability criterion threshold in the numerical model of the accumulation body.

[0030] Furthermore, when the measured water pressure value in the water pressure data of the post-disaster deposit to be collected and / or the measured deformation rate in the surrounding rock deformation data exceeds the preset physical threshold, a correction mechanism is automatically triggered. The correction mechanism is configured to dynamically adjust the permeability-related parameters or rheology-related parameters of the simulation configuration parameter set according to the risk level parameter.

[0031] Furthermore, the control command parameters of the equipment include: The control signal is at least one of the following: the operating permit status of the slag removal machinery, the start / stop status of the grouting equipment, and the trigger status of the audible and visual alarm device.

[0032] like Figure 3As shown, the present invention also provides a comprehensive evaluation and emergency response system for post-disaster deposits in underground engineering to implement the above-mentioned method. This system includes a data acquisition unit 1, an edge computing unit 2, a control execution unit 3, and a communication unit 4. The data acquisition unit 1 includes a 3D laser scanner, an industrial camera, a water pressure sensor, and a convergence meter installed at the site to be collected in the post-disaster deposit. The data acquisition unit 1 is configured to acquire at least one of the following in real time: 3D point cloud data, on-site image data, water pressure data, and surrounding rock deformation data of the site to be collected in the post-disaster deposit. The edge computing unit 2 is connected to the data acquisition unit 1 and includes a geological database 21, a point cloud processing module 22, a material identification module 23, and a mechanism parameter generation module 24. The system includes a simulation configuration parameter generation module 25 and a multi-field coupled simulation module 26; a geological database 21 configured to contain stratigraphic genetic information of tunnels at monitoring nodes; a point cloud processing module 22 configured to preprocess at least one set of acquired 3D point cloud data to construct a local coordinate system, and use the preprocessed 3D point cloud data in the local coordinate system to calculate the geometric morphological parameters of the post-disaster deposit to be collected; and a material identification module 23 configured to input the acquired field image data into a pre-trained convolutional neural network model, which is configured to obtain the probability value corresponding to each material category based on the input field image data, and determine the material category with the highest probability value as the post-disaster deposit to be collected. The system outputs material composition parameters of the affected area; the mechanism parameter generation module 24 is configured to: call the geological database 21 pre-stored in the edge computing terminal, read the stratigraphic genesis information of the tunnel corresponding to the mileage of the area to be collected in the post-disaster deposit, and combine the geometric morphology parameters and material composition parameters to generate a set of mechanism parameters characterizing the disaster formation mechanism; the simulation configuration parameter generation module 25 is configured to: automatically generate a simulation configuration parameter set based on the mechanism parameter set; the multi-field coupling simulation module 26 is configured to: input the simulation configuration parameter set to drive the numerical simulation engine to build a numerical model of the deposit, and calculate the fluid-structure interaction values ​​of the numerical model of the deposit to obtain the distribution characteristics of the displacement field, stress field and seepage field of the area to be collected in the post-disaster deposit; and from the post-disaster deposit... Key physical quantities are extracted from the distribution characteristics of the displacement field, stress field, and seepage field of the area to be collected and compared with preset physical threshold conditions to generate instability mode parameters and risk level parameters with correlation to characterize the instability characteristics of the accumulation body. The control execution unit 3 includes a programmable logic controller (PLC) and an actuator drive module 32. The PLC is configured to automatically generate control command parameters for the corresponding equipment based on the instability mode parameters and risk level parameters. The actuator drive module 32 is configured to receive control command parameters issued by the PLC to control the corresponding equipment to execute corresponding emergency measures. For example, the control signal is output by the PLC to realize the start-stop operation or access control of the slag removal machinery, grouting equipment, and audible and visual alarm devices.Communication unit 4 is used to establish a communication link between edge computing unit 2 and control execution unit 3 via industrial Ethernet or wireless communication module.

[0033] In some embodiments, the three-dimensional spatial information of the post-disaster debris is obtained through terrestrial three-dimensional laser scanning. The scanning equipment can be a phase-type or time-of-flight laser scanner with dustproof, moisture-proof, and low-light adaptability capabilities, such as the Faro Focus S350 or Leica BLK360.

[0034] The preferred scanning deployment method is: Three to five scanning stations are spaced at intervals along the tunnel culvert axis, with a preferred spacing of 5 to 8 meters between adjacent stations. Additional scanning stations are added in areas of abrupt geometric changes in the accumulation mass (including but not limited to steps, steep slopes, and areas of concentrated seepage). To ensure the accuracy of multi-station point cloud stitching, it is preferable to deploy 4 to 6 spherical or cylindrical targets as common control points within the accumulation area, with a preferred target diameter of 40 to 60 mm.

[0035] In this embodiment, the scanning parameters can be set as follows: Point cloud density not less than 80–120 points / cm 2 The ranging accuracy is no greater than 2–3 mm (@10 m), and the scanning time for each scanning station does not exceed 10 minutes. After multi-station registration, the overall point cloud registration error is preferably controlled within 5 mm.

[0036] Meanwhile, industrial-grade explosion-proof cameras are used to acquire surface images of the accumulation body, with an image resolution preferably not less than 4000×3000 pixels, covering the entire accumulation body and key detailed areas. Environmental parameters such as water pressure, surrounding rock convergence or settlement are collected synchronously through water pressure sensors and convergence meters, and transmitted in real time to the edge computing terminal via wired or wireless means.

[0037] In some implementations, the edge computing terminal sequentially performs statistical outlier removal and radius filtering noise reduction on the registered 3D point cloud data, and establishes a local coordinate system, in which the tunnel culvert axis is defined as the X-axis, the vertical axis as the Z-axis, the horizontal axis as the Y-axis, and the lowest point of the accumulation body as the origin of the coordinate system.

[0038] In the above, the system automatically calculates the following geometric parameters: The stack height H and maximum width W; The natural angle of repose α fitted along the longitudinal section of the accumulator; Porosity φ calculated by voxelization (voxel side length preferably 1 cm); The surface fractal dimension D was calculated using the box counting method. The number of steps N and the height of a single step hᵢ are counted along the vertical direction; The number of bedding planes M and the dip angle θ are obtained by normal vector clustering and plane fitting.

[0039] In a preferred discrimination rule, morphological features are automatically determined according to the following progressive conditions: When the height-to-width ratio H / W of the accumulation is in the range of 0.3–0.5 and the natural angle of repose α is in the range of 30°–40°, it is determined to be a conical structure; When the number of steps N≥3 and the height hᵢ of a single step is in the range of 5–20cm, it is determined to be a stepped structure. When the number of bedding planes M≥2 and the dip angle θ≤10° of the bedding planes, it is determined to be a layered structure; When the porosity φ of the packing is ≥25% and the fractal dimension D is in the range of 1.2–1.8, it is determined to be an irregular structure; When the number of morphological partitions K of the accumulation body is greater than or equal to 2, and the area of ​​the largest partition does not exceed 70%, it is determined to be a composite structure.

[0040] In a preferred embodiment, the discrimination result of the morphological feature is used to automatically determine the partitioning method, mesh type and mesh density of the initial geometric model, and to determine the location range of potential contact surfaces, slip surfaces or weak surfaces in the simulation model, so as to improve the adaptability of numerical simulation to the real geometric shape of the accumulation body.

[0041] In a preferred embodiment, the material composition of the accumulation body is determined by image recognition based on deep learning. The pre-trained convolutional neural network model can be an EfficientNet, ResNet, or MobileNet architecture, and is pre-trained on a dataset containing images of various underground engineering accumulation bodies.

[0042] Specifically, the training dataset used for pre-training the convolutional neural network model contains no fewer than 10,000 labeled samples, and the material categories include, but are not limited to: soil, sand, gravel, debris flow mixture, mudflow, rock fragments, gold-bearing structures, etc.

[0043] In the above, the probability distribution of each material category is output by a pre-trained convolutional neural network model. Finally, the system selects the material category with the highest probability as the main material composition parameter of the accumulation, and records the second highest probability category as an auxiliary criterion to identify foreign matter or multiphase mixed structures.

[0044] In a preferred embodiment, the system obtains strength characteristics, permeability characteristics, rheological characteristics and structural characteristics parameters based on the material composition parameters and the geological database 21, and calculates the risk level parameter S in a weighted manner. The value of S is obtained by weighting multiple physical characteristic parameters and is used for the dynamic correction of subsequent simulation configuration parameters and control thresholds.

[0045] In the above, the weights are configured as follows: The strength characteristic weight is 25%, and it is the average value of the cohesion, internal friction angle and compressive strength corresponding to the grade. The permeability characteristic weight is 20%, and it is obtained by weighting the permeability coefficient (weight 60%) and the measured water pressure level (weight 40%). The rheological property weight is 25%, which is determined by the level corresponding to the measured deformation rate. The structural characteristics have a weight of 30%, which is the average of porosity and density levels.

[0046] In a preferred dynamic adjustment mechanism, when the measured water pressure is not less than 0.5 MPa, the system automatically increases the weight of permeability characteristics by 5 percentage points; when the measured deformation rate is not less than 0.8 mm / d, the system automatically increases the weight of rheological characteristics by 5 percentage points and correspondingly decreases the weights of other characteristics to keep the total weight of 100%.

[0047] It should be noted that the above risk level parameters are not used as independent evaluation conclusions, but as adjustment factors for material constitutive parameter reduction coefficients, boundary load amplitudes, and simulation safety factors.

[0048] In a preferred embodiment, when the risk level parameter S meets the preset physical triggering conditions, the system automatically dynamically corrects the simulation configuration parameter set and calls the integrated numerical simulation engine to perform multi-field coupling analysis.

[0049] Preferably, the point cloud data is reconstructed into an STL mesh model as the initial geometric model; the material constitutive model is automatically matched based on the material composition parameters, for example: Soil or debris flow type deposits are matched with the Mohr-Coulomb model and superimposed with creep or consolidation parameters; fractured rock block type deposits are matched with elastoplastic or damage models.

[0050] Boundary conditions are applied based on genetic mechanism parameters, such as applying lateral high-stress boundaries to soft rock types with high geostress and applying pore water pressure boundaries to water-rich fault types.

[0051] After comparing the key physical quantities in the distribution characteristics of the displacement field, stress field, and seepage field output by the simulation with the preset physical thresholds, instability mode parameters (such as flow instability, sliding instability, collapse instability, interface instability, or combined instability) and risk level parameters are generated.

[0052] In a preferred control logic, such as Figure 2As shown, different combinations of equipment control signals correspond to different instability mode parameters and risk level parameters. These signals are used to trigger audible and visual alarms, restrict or release mechanical operation permissions, and start grouting or reinforcement equipment. This enables automated emergency response to post-disaster debris, eliminating the need for manual on-site judgment and operation. It not only avoids the subjectivity and experience limitations of human decision-making but also accurately matches response measures according to the risk level of different areas of the debris, such as differentiated grouting intensity and targeted slag removal range. This avoids over- or under-response issues, improves emergency response efficiency and resource utilization efficiency, and ensures the operational safety of rescue personnel.

[0053] In this embodiment, a test scenario is constructed using the post-disaster comprehensive evaluation and emergency response to a sudden water and mudslide in a mountain tunnel traversing a water-rich fault zone. The test scenario involves a highway tunnel traversing the F3 regional fault fracture zone at a depth of approximately 320 m. During construction at kilometer marker ZK45+780, a sudden high-pressure water inrush occurred at the tunnel face, carrying mud and sand into the tunnel. Within 30 minutes, the accumulated mass reached a length of 18 m, and the peak water inrush volume was approximately 120 m³. 3 / h, low visibility, and partial collapse of the initial support. It is necessary to quickly obtain the geometry and hydraulic state of the accumulation mass under high-risk conditions and generate executable disposal control command parameters to ensure the safety of personnel and equipment. Therefore, based on the comprehensive evaluation and emergency disposal method for post-disaster accumulation masses in underground engineering provided in this application, corresponding control command parameters are configured to drive the actuator drive module 32 to execute corresponding emergency measures, including the following steps: Step 1, Data Collection: (1) 3D scanning: Three scanning stations were set up using a dustproof Leica BLK360 scanner (with a spacing of 6m between adjacent scanning stations), focusing on covering the front of the deposition and the concentrated seepage area; four waterproof target spheres (50mm in diameter) were also set up. The point cloud density was approximately 98 points / cm². 2 The registration error is ≤4.8mm.

[0054] (2) Image and environmental parameters: Eight images were acquired using an explosion-proof infrared enhanced industrial camera (which can penetrate water mist). The water pressure measured by the water pressure sensor was 0.68 MPa, and the corresponding deformation rate for convergence / settlement monitoring was 1.25 mm / d.

[0055] Step 2, Feature parsing and parameter set generation: (1) Point cloud processing and geometric morphology parameters: outlier removal and filtering are performed on the point cloud, a local coordinate system is established, and geometric features such as the aspect ratio H / W of the accumulation body are 0.41, the angle of repose α is 35°, the porosity φ is 42%, the fractal dimension D is 1.35, and the number of steps N is 0 are calculated.

[0056] (2) Material composition parameters: The image was input into the pre-trained EfficientNet-B4 model, and the output probability distribution was: 52% saturated muddy sand, 30% gravel, 10% soluble salt crystals, and 8% rock powder. Based on the maximum probability, it was determined to be a mixture of high water content sand / gravel and recorded as "high permeability saturated granular material" material parameters.

[0057] (3) Mechanism parameter set: The geological database 21 is called to read the genetic information of the mileage, which is located in the F3 fault fracture zone and has strong water conductivity, and form a mechanism parameter set containing fields such as water-conducting structure, degree of fracture, and possible seepage channels.

[0058] Step 3: Simulation configuration parameter set and multi-field coupled simulation: (1) Geometric model parameters: The registered point cloud is reconstructed into an STL mesh model, and the main geometric scales (such as H=5.2m, W=12.6m) are extracted as geometric model parameters.

[0059] (2) Material constitutive model parameter selection: Match the Mohr-Coulomb model from the material model library according to the material composition parameters and superimpose consolidation / seepage related parameters (such as Biot consolidation or seepage coupling parameters), and read the corresponding parameter range.

[0060] (3) Boundary load configuration parameters: The lateral water pressure boundary (0.68MPa) and bottom permeability / drainage conditions are loaded according to the mechanism parameter set, and the initial pore water pressure field is set to static water distribution.

[0061] (4) Call the simulation engine: Perform fluid-structure interaction calculations through the FLAC3D API and output the distribution characteristics of displacement field, stress field and seepage field.

[0062] Step 4, Instability Feature Identification and Threshold Triggering: Key physical quantities are extracted from the distribution characteristics of the displacement, stress, and seepage fields output by the simulation, such as: the expansion distance of the accumulation front, surface velocity, cumulative outflow volume, displacement rate, and pore pressure change amplitude within 24 hours. These key physical quantities are then compared with preset physical threshold conditions to obtain instability mode parameters and risk level parameters.

[0063] In this embodiment, the simulation output shows: the maximum extension length is approximately 28m, the maximum surface flow velocity is approximately 420mm / s, and the cumulative outflow volume is approximately 180m³. 3 If the preset "high-risk flow instability" threshold range is met, the corresponding instability mode parameters (flow instability) and risk level parameters (high risk) are generated.

[0064] Step 5: Control the linkage between command parameter generation and execution: The edge computing terminal generates equipment control command parameters based on instability mode parameters and risk level parameters, and pushes them to the BIM emergency platform via API interface or sends them directly to the PLC controller to drive the control execution unit 3 to output control signals, for example: Trigger the audible and visual alarm device and broadcast an evacuation order; Lock down the operating permissions of the slag removal machinery inside the tunnel and cut off the power supply to non-essential equipment. Start the remote grouting equipment or issue a grouting plan trigger signal; When the water pressure drops below the safety threshold and the displacement rate decreases, the lock signal of some equipment is released and phased resumption of work is allowed.

[0065] In this embodiment, the system outputs a combination of high-risk control signals: triggering the corresponding alarm control signal combination and personnel evacuation signal, locking the mechanical operation permission signal in the tunnel culvert, pushing the surface grouting treatment trigger signal, and continuously monitoring water pressure and displacement changes until the water pressure drops to a preset safety threshold, such as below 0.1MPa, before allowing the resumption of work assessment process.

Claims

1. A comprehensive evaluation and emergency response method for post-disaster debris accumulation in underground engineering projects, characterized in that, The method includes the following steps: A geological database (21) is generated by acquiring stratigraphic genetic information of the tunnel at the node to be monitored, and pre-stored in an edge computing terminal deployed at the tunnel site. The edge computing terminal automatically executes the emergency response procedure according to the set control program. The emergency response procedure is set as follows: Real-time acquisition of at least one of the following: three-dimensional point cloud data, on-site image data, water pressure data, and surrounding rock deformation data of the post-disaster debris body to be collected. After preprocessing at least one set of the acquired three-dimensional point cloud data, a local coordinate system is constructed, and the geometric morphological parameters of the part of the post-disaster accumulation body to be collected are calculated using the preprocessed three-dimensional point cloud data in the local coordinate system. The acquired on-site image data is input into a pre-trained convolutional neural network model. The pre-trained convolutional neural network model is configured to: obtain the probability value corresponding to each material category based on the input on-site image data, and determine the material category with the highest probability value as the material composition parameter of the part to be collected from the post-disaster deposit and output it. The geological database (21) pre-stored in the edge computing terminal is called to read the stratigraphic genesis information of the tunnel corresponding to the mileage of the post-disaster deposit to be collected, and the geometric morphology parameters and material composition parameters are combined to generate a set of mechanism parameters characterizing the disaster formation mechanism. Based on the aforementioned mechanism parameter set, a simulation configuration parameter set is automatically generated to drive the numerical simulation engine to construct a numerical model of the accumulation body, and the fluid-structure interaction values ​​of the numerical model of the accumulation body are calculated to obtain the distribution characteristics of the displacement field, stress field and seepage field of the part of the post-disaster accumulation body to be collected. Key physical quantities are extracted from the distribution characteristics of displacement field, stress field and seepage field of the post-disaster deposit to be collected and compared with preset physical threshold conditions to generate instability mode parameters and risk level parameters with correlation to characterize the instability characteristics of the deposit. Based on the instability mode parameters and the risk level parameters, control command parameters for the corresponding devices are automatically generated and sent to the corresponding devices through the communication interface to drive them to execute the corresponding emergency measures.

2. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The method for preprocessing the 3D point cloud data specifically includes: The acquired 3D point cloud data is subjected to statistical filtering and multi-site registration processing.

3. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The simulation configuration parameter set includes at least geometric model parameters, material constitutive model selection parameters, and boundary load configuration parameters.

4. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The simulation configuration parameter set is set as the input interface parameter of the numerical simulation engine, which is used to trigger the numerical simulation engine to retrieve the corresponding material constitutive model data, boundary loads and initial conditions related to the cause of disasters, and the construction constraint parameters of the numerical model of the accumulation body from the preset configuration file.

5. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 4, characterized in that, The construction constraint parameters of the numerical model of the accumulation body include at least one of the following: calculation step size, convergence criterion, and safety factor.

6. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The geometric modeling method of the numerical model of the accumulation body is adapted and selected based on the determination results of the geometric morphology parameters, which include: Determine the partitioning method of the initial geometric model of the accumulation body, determine the mesh type or mesh density of different regions, and determine at least one of the following: the location and range of the contact surface, slip surface or weak surface in the simulation model.

7. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The reduction factor of the material constitutive model selection parameter, the loading amplitude or loading rate of the boundary load configuration parameter, and the safety factor or stability criterion threshold in the simulation calculation process are calibrated in real time according to the risk level parameter.

8. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, When the measured water pressure value in the water pressure data of the post-disaster deposit to be collected and / or the measured deformation rate in the surrounding rock deformation data exceed the preset physical threshold, a correction mechanism is automatically triggered. The correction mechanism is configured to dynamically adjust the permeability-related parameters or rheology-related parameters of the simulation configuration parameter set according to the risk level parameter.

9. The method for comprehensive evaluation and emergency response of post-disaster debris in underground engineering projects according to claim 1, characterized in that, The control command parameters of the device include: The control signal is at least one of the following: the operating permit status of the slag removal machinery, the start / stop status of the grouting equipment, and the trigger status of the audible and visual alarm device.

10. A comprehensive evaluation and emergency response system for post-disaster debris in underground engineering projects, used to implement the method described in any one of claims 1-9, characterized in that, The system includes: The data acquisition unit (1) is configured to acquire at least one of the following in real time: three-dimensional point cloud data, on-site image data, water pressure data, and surrounding rock deformation data of the post-disaster deposit to be collected. An edge computing unit (2) is connected to the data acquisition unit (1). The edge computing unit (2) includes a geological database (21), a point cloud processing module (22), a material identification module (23), a mechanism parameter generation module (24), a simulation configuration parameter generation module (25), and a multi-field coupling simulation module (26). The geological database (21) is configured to have built-in stratigraphic genetic information of the tunnel on the monitoring node. The point cloud processing module (22) is configured to preprocess at least one set of the acquired three-dimensional point cloud data to construct a local coordinate system, and use the preprocessed three-dimensional point cloud data in the local coordinate system to calculate the geometric morphological parameters of the post-disaster deposit to be collected. The material identification module (23) is configured to input the acquired field image data into a pre-trained convolutional neural network model. The pre-trained convolutional neural network model is configured to obtain the probability value corresponding to each material category according to the input field image data, and determine the material category with the highest probability value as the post-disaster deposit. The material composition parameters of the part to be collected in the accumulation are output; the mechanism parameter generation module (24) is configured to: call the geological database (21) pre-stored in the edge computing terminal, read the stratigraphic genesis information of the tunnel corresponding to the mileage of the part to be collected in the post-disaster accumulation, and combine the geometric morphology parameters and material composition parameters to generate a mechanism parameter set characterizing the disaster formation mechanism; the simulation configuration parameter generation module (25) is configured to: automatically generate a simulation configuration parameter set based on the mechanism parameter set; the multi-field coupling simulation module (26) is configured to: input the simulation configuration parameter set to drive the numerical simulation engine to build a numerical model of the accumulation, and calculate the fluid-structure coupling value of the numerical model of the accumulation to obtain the distribution characteristics of the displacement field, stress field and seepage field of the part to be collected in the post-disaster accumulation; and extract key physical quantities from the distribution characteristics of the displacement field, stress field and seepage field of the part to be collected in the post-disaster accumulation and compare them with the preset physical threshold conditions to generate instability mode parameters and risk degree parameters with correlation to characterize the instability characteristics of the accumulation; The control execution unit (3) includes a programmable logic controller and an actuator drive module (32). The programmable logic controller is configured to automatically generate control command parameters for the corresponding device based on the instability mode parameters and the risk level parameters. The actuator drive module (32) is configured to receive the control command parameters issued by the programmable logic controller to control the corresponding device to execute the corresponding emergency measures. The communication unit (4) is used to establish a communication link between the edge computing unit (2) and the control execution unit (3) via an industrial Ethernet or wireless communication module.

Images