Underground roadway multi-node water burst disaster emergency decision-making method and system

Abstract

The invention belongs to the related technical field of underground engineering safety, and provides an underground roadway multi-node water gushing disaster emergency decision-making method and system, which comprehensively considers water gushing pressure, roadway resistance and fluid characteristics, establishes a water gushing forward arrival time prediction model, and can control the prediction error within + / -15%; establishing a collaborative optimization framework of an evacuation path and a plugging strategy, bringing personnel safe evacuation and plugging device energy consumption into a unified decision-making system, and dynamically balancing personnel safety and plugging device energy consumption; a personnel physical power index is introduced into the improved A * algorithm, the evacuation priority is dynamically adjusted, the water burst forward arrival time margin and the personnel physical power index are added into a path cost function to optimize weak personnel path selection, and the overall emergency response efficiency is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to underground engineering safety, and in particular relates to an emergency decision-making method and system for multi-node water inrush disasters in underground tunnels. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In recent years, with the increasing sophistication of above-ground infrastructure, geotechnical engineering has gradually expanded into underground space development, shifting the project implementation model from rapid growth to a quality and efficiency-oriented approach. Against this backdrop, underground space development projects involving coal mine excavation, tunneling, and underground construction are becoming increasingly frequent. However, emergency response to sudden water inrush disasters still faces a bottleneck due to insufficient intelligence.

[0004] Specifically, as a typical underground transportation network, the spatial structure of underground tunnel systems is highly complex, exhibiting a multi-level bifurcation topological characteristic. When a water inrush occurs, the dynamic propagation of water flow exhibits significant nonlinear diffusion characteristics, influenced by the combined effects of geological structural differences, construction disturbance effects, and the three-dimensional spatial topological structure of the tunnel network. Especially at tunnel bifurcations, water inrush is affected by the multi-physics coupling of gravitational potential energy gradients, tunnel inclination distribution, and local water pressure differences, resulting in multi-path diversion. The velocity field distribution and front propulsion rate of each branched flow show significant spatial heterogeneity.

[0005] However, existing technical solutions are mostly based on manual experience and static contingency plans, relying on analogical deductions from limited historical cases or simple extrapolation of local monitoring data. These solutions present significant limitations when addressing concurrent water inrush at multiple nodes in complex networked tunnel systems. Traditional methodologies fail to effectively integrate tunnel three-dimensional spatial morphological parameters, water inrush diffusion dynamics models, and personnel movement status monitoring data, resulting in insufficiently accurate predictions of disaster evolution. In particular, with regard to dynamic coordination mechanisms, existing systems lack the collaborative decision-making capabilities to optimize the activation timing of plugging devices and adjust evacuation routes in real time, which can easily lead to misallocation of emergency resources. Specifically, this can manifest as delayed plugging triggering secondary disaster risks or redundant protective measures leading to wasted resources. Furthermore, existing technical frameworks fail to fully consider the coupling relationship between the three-dimensional spatial properties of tunnels and the dynamic diffusion characteristics of water inrush, nor do they establish a spatiotemporal mapping mechanism between the water inrush front propulsion model and the personnel evacuation velocity field, making it difficult to achieve real-time prediction and dynamic control of water inrush diffusion routes. Summary of the Invention

[0006] In order to overcome the shortcomings of the above-mentioned existing technologies, the present invention provides an emergency decision-making method and system for multi-node water inrush disasters in underground tunnels, an intelligent decision-making system that dynamically generates optimal evacuation paths and blocking strategies, realizes dynamic regulation of blocking control, and significantly improves the overall emergency response efficiency.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides an emergency decision-making method for multi-node water inrush disasters in underground tunnels, comprising: Based on the tunnel BIM model, tunnel water inrush parameters and personnel status, a water inrush diffusion prediction model is constructed. The diffusion path of water inrush in different branch tunnels and the time when the water inrush front reaches each tunnel node are predicted based on the water inrush diffusion prediction model. An evacuation-plugging collaborative optimization model was constructed based on personnel safety and the energy consumption of the plugging device. An improved A* algorithm was used to generate a safe evacuation path for personnel based on the water inrush prediction results. With maximizing personnel safety and minimizing the energy consumption of the plugging device as constraints, a safe evacuation plan for personnel and an optimal plugging plan were obtained. The improved A* algorithm introduced the personnel physical strength index to dynamically adjust the evacuation priority, and the arrival time margin of the water inrush front was added to the path cost function.

[0008] In a second aspect, the present invention provides an underground tunnel multi-node water inrush disaster emergency decision-making system, comprising: The prediction module is configured to: construct a water inrush diffusion prediction model based on the roadway BIM model, roadway water inrush parameters, and personnel status; and predict the diffusion path of water inrush in different branch roadways and the time when the water inrush front arrives at each roadway node based on the water inrush diffusion prediction model; The decision-making module is configured to: construct an evacuation-blocking collaborative optimization model based on personnel safety and the energy consumption of the blocking device; use the improved A* algorithm to generate a safe evacuation path for personnel based on the water inrush prediction results; and use maximizing personnel safety and minimizing the energy consumption of the blocking device as constraints to obtain a safe evacuation plan for personnel and an optimal blocking plan; among them, the improved A* algorithm introduces the personnel physical index to dynamically adjust the evacuation priority; and adds the arrival time margin of the water inrush front to the path cost function.

[0009] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0010] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0011] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which implements the method described in the first aspect when executed by a processor.

[0012] One or more of the above technical solutions have the following beneficial effects: In the present invention, a water surge front arrival time prediction model is established by comprehensively considering the water surge pressure, tunnel resistance and fluid characteristics, and the prediction error can be controlled within ±15%; a collaborative optimization framework for evacuation paths and blocking strategies is established, and the safe evacuation of personnel and the energy consumption of blocking devices are incorporated into a unified decision-making system to dynamically balance personnel safety and the energy consumption of blocking devices; the improved A* algorithm introduces the personnel physical strength index, dynamically adjusts the evacuation priority, and adds the water surge front arrival time margin to the path cost function. The personnel physical strength index optimizes the path selection of weak personnel, significantly improving the overall emergency response efficiency.

[0013] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0015] Figure 1 This is an overall block diagram of emergency decision-making for multi-node water inrush disasters in underground tunnels in the first embodiment of the present invention; Figure 2 This is a flow chart of the water diffusion prediction model in Example 1 of the present invention; Figure 3 This is a flow chart of the evacuation-blocking collaborative optimization algorithm in Example 1 of the present invention; Figure 4 This is a pressure-time curve diagram in Example 1 of the present invention. DETAILED DESCRIPTION

[0016] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0017] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0018] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0019] Example 1 This embodiment discloses an emergency decision-making method for multi-node water inrush disasters in underground tunnels, including: Based on the tunnel BIM model, tunnel water inrush parameters and personnel status, a water inrush diffusion prediction model is constructed. The diffusion path of water inrush in different branch tunnels and the time when the water inrush front reaches each tunnel node are predicted based on the water inrush diffusion prediction model. An improved A* algorithm is used to generate safe evacuation paths for personnel based on the water inrush prediction results. An evacuation-blocking collaborative optimization model is constructed according to the personnel safety and the energy consumption of the blocking device. With maximizing personnel safety and minimizing the energy consumption of the blocking device as constraints, a safe evacuation plan for personnel and an optimal blocking plan are obtained. The improved A* algorithm introduces the personnel physical strength index to dynamically adjust the evacuation priority, and the arrival time margin of the water inrush front is added to the path cost function.

[0020] This embodiment provides an intelligent decision-making system that can integrate personnel locations, water gushing parameters and tunnel structure data in real time to dynamically generate optimal evacuation paths and blocking strategies, thereby achieving real-time prediction of water gushing diffusion paths and dynamic regulation of blocking control.

[0021] The following combination Figure 1 The emergency decision-making method for multi-node water inrush disaster in underground tunnels proposed in this embodiment is described in detail: Step 1: Based on the tunnel BIM model, tunnel water inrush parameters and personnel status, a water inrush diffusion prediction model is constructed. Based on the water inrush diffusion prediction model, the diffusion path of water inrush in different branch tunnels and the time when the water inrush front arrives at each tunnel node are predicted.

[0022] The water inrush spread prediction model includes a water inrush front arrival time prediction model and a water inrush spread path dynamic prediction model. The water inrush front arrival time prediction model takes into account water inrush pressure, tunnel resistance, and fluid characteristics. During the water inrush arrival time prediction process, the arrival time and location of the water inrush are predicted, and this is dynamically updated by updating the water inrush parameters.

[0023] Dynamic prediction model of water gushing diffusion path: Through real-time sensor data (such as water pressure and flow velocity) and the three-dimensional topological structure of the tunnel, the water gushing velocity field distribution is dynamically corrected to predict the diffusion path of water gushing in different branch tunnels.

[0024] A distributed sensor network is deployed at the intersection of alleys, such as multi-parameter sensors, integrated temperature and humidity sensors, CO concentration sensors and pressure sensors, with a data update frequency of ≥10Hz, to obtain real-time data of the alleys; the real-time status of personnel can be obtained by having personnel wearing UWB positioning helmets.

[0025] A water inrush diffusion prediction model was constructed based on the tunnel BIM model and real-time tunnel data. The water inrush pressure, tunnel resistance, and fluid characteristics were comprehensively considered. The Navier-Stokes coupled porous media seepage theory equation was simplified through dimensional analysis to establish the arrival time of the water inrush front. computational model.

[0026] The Navier-Stokes equations are simplified to:

[0027] Where ν is the kinematic viscosity of the fluid, θ is the inclination angle between the flow direction and the horizontal plane, combined with the Forchheimer seepage formula

[0028] The dimensionless water front velocity is obtained through dimensional analysis:

[0029] Where k is the permeability of the porous medium, μ is the dynamic viscosity of the fluid, ∇P is the pressure gradient, is the dimensionless drag coefficient, is the characteristic length, such as the flow path length, unit: m, is the dimensionless pressure difference.

[0030] Arrival time of water front Prediction model:

[0031] Where L is the tunnel length, ΔP is the pressure difference between the water inflow point and the target position (monitored in real time by the sensor), ρ is the fluid density, and λ is the tunnel resistance coefficient.

[0032] Step 2: An evacuation-blocking collaborative optimization model is constructed based on personnel safety and the energy consumption of the blocking device. An improved A* algorithm is used to generate a safe evacuation path for personnel based on the water inrush prediction results. With maximizing personnel safety and minimizing the energy consumption of the blocking device as constraints, a safe evacuation plan for personnel and an optimal blocking plan are obtained. The improved A* algorithm introduces the personnel physical strength index to dynamically adjust the evacuation priority, and the arrival time margin of the water inrush front is added to the path cost function.

[0033] The improved A* algorithm is used to generate safe evacuation paths for personnel based on the water inrush prediction results, and the shortest safe evacuation path for personnel is generated, specifically: Step 21: Algorithm initialization.

[0034] Input parameter settings: tunnel BIM model, water front arrival time prediction value (Tarrive), personnel real-time location and physiological parameters (moving speed vi, heart rate HRi, etc.), and blocking device status.

[0035] Initialize the physical sensitivity coefficient (α, ranging from 0.5 to 1.2) and the risk aversion coefficient (β, ranging from 1.5 to 3.0).

[0036] Data structure initialization: Add the starting point to the open list and set its actual cost g(n)=0; Initialize the heuristic cost h(n); Dynamic parameter binding: Receive sensor data (water pressure, flow rate, roadway resistance coefficient) in real time and update Tarrive; Step 22: Path search loop.

[0037] Select current node: select total cost from open list f(n)=g(n)+h(n) The smallest node, n, is the current expansion node.

[0038] Target node determination: If the current point is the target point, that is, the safe area, backtrack to the parent node to generate a complete path and end the search; otherwise, remove node n from the open list and add it to the closed list.

[0039] Traverse adjacent nodes: Perform the following operations on all adjacent nodes m of node n: Skip closed list nodes: If m is already in the closed list, ignore it; Calculate the actual cost g(m) of the path from the starting point to node m; estimate the evacuation time based on the real-time speed of the personnel.

[0040] Improved A* algorithm heuristic function is modified to:

[0041] in, is the physical sensitivity coefficient, which is used to adjust the influence of physical strength on path selection, with a value range of 0.5-1.2; f(n) is the estimated total cost of node n; g(n) is the actual cost from the starting point to node n, and its unit is related to the path length, such as meter; h(n) is the heuristic cost from node n to the target point, and its unit is related to the path length, such as meter; α is the physical sensitivity coefficient, which is dimensionless and has a value range of 0.5-1.2; Si is the physical index, which is dimensionless and reflects the individual's real-time physical status, and its value range is usually 0-1.

[0042] Physical fitness index :

[0043] in, is the real-time moving speed (unit: m / s); is the maximum movement speed of the individual, unit: m / s; Real-time heart rate, unit: beats / minute, bpm; is the individual's maximum heart rate, unit: bpm; is the resting heart rate, unit: bpm.

[0044] in, is the real-time moving speed, is the heart rate, For resting heart rate, dynamic weight adjustment is performed and the heuristic function is modified.

[0045] The risk aversion mechanism is to add the arrival time margin of the water front to the path cost function:

[0046] in, is the risk aversion coefficient, which is experimentally determined to be 1.5-3.0. is the physical length of the path, in meters; β is the risk aversion coefficient (dimensionless), with an experimental value of 1.5-3.0, which is used to amplify the impact of risk on cost; The predicted time for the water front to reach the current position, unit: seconds, s; The estimated evacuation time from the current location to the safe point, in seconds.

[0047] Taking the evacuation time, energy consumption of the blocking device and system response delay as constraints, an evacuation-blocking collaborative optimization model is constructed. The system response delay is the activation time of each blocking device. Control and adjust the start time according to personnel safety.

[0048] like Figure 3-Figure 4 As shown in the figure, the evacuation-blocking collaborative optimization model takes the blocking device start-up time tblock as the decision variable and establishes the constraint conditions:

[0049] in, For the i Personnel safety weight, For the i The distance between personnel and the front of the water surge, is the global energy consumption penalty coefficient, is the step energy consumption coefficient, For the j The activation time of each blocking device, n is the number of personnel,m is the number of occluding devices.

[0050] Indicates personnel safety; Indicates the energy consumption penalty of the blocking device, which is used to determine the start-up time of the blocking device and control energy consumption. When the personnel safety is low, the blocking device is started quickly to reduce energy consumption. When the personnel safety is high, the device start-up speed is reduced in consideration of energy consumption.

[0051] When the plugging device needs to be activated quickly and preferentially, the activation priority and time window of the plugging device are dynamically adjusted through the reinforcement learning algorithm (PPO). The reinforcement learning algorithm adopts a hybrid optimization strategy, namely a hybrid algorithm of genetic algorithm (GA) + particle swarm optimization (PSO). Among them, GA is responsible for global search (population size 50, crossover probability 0.8), and PSO handles local extreme values ​​(inertia weight linearly decreases from 0.9 to 0.4). The convergence speed is 60% faster than that of a single algorithm.

[0052] Specifically, the parameters of each plugging device's start time, start queue order, interval time between adjacent plugging devices, and energy consumption coefficient are packaged into a "scheduling plan package". , such as: [150 seconds to start, gate 3 priority, interval 25 seconds, energy consumption coefficient 0.8].

[0053] Global initial screening (GA stage): Population size 50, using tournament selection, size = 5; Crossover operation: OX crossover was used with a probability of 0.8 and arithmetic crossover of continuous parameters; Mutation operation: perform Gaussian mutation on the interval Δt (σ=5 seconds), Randomly swap two positions; The computer randomly generates 50 different "scheduling plan packages", evaluates each plan's personnel safety, energy consumption and other indicators, and eliminates plans with poor results.

[0054] Perform "genetic crossover" on excellent solutions, such as combining the time parameters of solution A with the gate sequence of solution B to generate a new generation of optimized solutions.

[0055] Local fine-tuning (PSO stage):

[0056] Among them, the subscript i represents the i-th particle, which represents a blocking solution, that is, the optimized solution obtained by global screening, such as a time window + priority combination. The time window determines when to start blocking, that is, the interval between adjacent blocking devices in the information encoding stage.

[0057] The subscript d is the dimension of the parameter, for example, the first dimension is the time window length , the second dimension is the priority sequence etc. The superscript k represents the number of previous iterations.

[0058] It represents the velocity of particle i in the dth dimension in the kth iteration, that is, the adjustment amount of the parameter, in seconds / iteration; for example, if the time window dimension d=1, This means that at the fifth iteration, the particle wants to extend the time window by 3 seconds.

[0059] Indicates the position of particle i in the dth dimension in the kth iteration, that is, the current value of the parameter. For example, if the time window dimension d=1, =150 means the current time window is set to 150 seconds. w is the inertia weight, which controls the inertia of the particle to maintain its original velocity. A large weight (such as 0.9) in the early stage allows for global exploration, while a small weight (such as 0.4) in the later stage focuses on local optimization.

[0060] 、 is the learning factor, which is set to 1.5 in this embodiment to control the influence intensity of individual experience and group experience respectively. 、 is a random number (0~1), and random perturbations are introduced to avoid falling into local optimality.

[0061] represents the best position of particle i in the dth dimension in history, that is, the individual optimal solution. If the time window of particle i has reached 132 seconds, then =132.

[0062] The optimal position of the dth dimension in the entire particle swarm is the optimal solution of the group. If the optimal time window of all particles is 128 seconds, then =128.

[0063] For the top 10% of the best-performing solutions, simulate "particle swarm flight": gradually adjust the time parameters, such as: 150 seconds → 148 seconds → 145 seconds, to find a more precise optimal start time.

[0064] The system automatically monitors the spread of gushing water. If it detects a sudden acceleration of the water flow (>0.3m / s²), it immediately increases the weight of the "personnel safety" indicator to 80%.

[0065] A multi-objective fitness function is established to calculate the corresponding score of the optimized blocking scheme, the blocking scheme is quantified, and the appropriate blocking scheme is selected according to the quantification result.

[0066] Multi-objective fitness function:

[0067] in, 、 、 are weight values, indicating the priority of different goals. 、 、 The sum is 1. The actual number of people safely evacuated; is the total number of people who need to be evacuated; : Total energy consumption of the plugging device (unit: kilojoule kJ); The larger the value, the higher the safety. For example, 0.95 indicates that 95% of personnel are safe. β is the energy consumption penalty coefficient. In this embodiment, it is set to 0.001 to adjust the impact of energy consumption on the score. Energy consumption is mapped to a value from 0 to 1. The lower the energy consumption, the higher the score. :Time window margin = blocking start time window - water front arrival time. For example: If the time window is set to 132 seconds, it takes 120 seconds for water to arrive, then =12 seconds. The maximum allowed time margin is used for normalization processing and is set to 60 seconds in this embodiment.

[0068] When the water surges faster, From 0.6→0.8 (safety first); under normal working conditions =0.6, =0.3, =0.1, for dynamic adjustment.

[0069] Dynamic adaptation: In the case of coal mine flooding, the system's initial plan was to "activate three gates in sequence after 150 seconds." After 15 rounds of optimization, it was adjusted to "prioritize the activation of Gate No. 1 at 132 seconds, shortening the interval to 18 seconds." This reduced the blocking energy consumption by 39% and provided an additional 23 seconds of safety time for personnel evacuation.

[0070] This embodiment optimizes computational complexity and adopts hierarchical path planning, i.e., coarse-grained global planning, to quickly generate an initial evacuation route, avoiding known high-risk areas such as downstream of a water gushing point and low-lying areas. Fine-grained local corrections are used to adjust path details in real time to avoid sudden obstacles such as localized water accumulation, operation of sealing devices, and changes in the dynamic risk field. This is equivalent to refining the model to grid units, updating the corresponding grids, using improved A* for path planning, dividing the roadway network into grid units (e.g., with a side length of 5m), building a dynamic update mechanism, and planning the trigger conditions for recalculating the route every 3 seconds.

[0071] In fine-grained local correction, the physical index (Si) and physical sensitivity coefficient (α) are used to dynamically modify the path cost function of the A* algorithm to achieve real-time optimization of the escape path of personnel. Specifically, the system collects the movement speed and heart rate in real time through the sensors worn by the personnel, and calculates the physical index by combining the preset individual maximum movement speed, maximum heart rate and resting heart rate.

[0072] in, is the real-time moving speed (unit: m / s); is the maximum movement speed of the individual, unit: m / s; Real-time heart rate, unit: beats / minute, bpm; is the individual's maximum heart rate, unit: bpm; is the resting heart rate, unit: bpm.

[0073] This index quantifies a person's real-time physical condition, with lower values ​​indicating greater physical exertion. Meanwhile, the physical sensitivity coefficient α (in this example, ranging from 0.5 to 1.2) is dynamically adjusted based on the severity of the disaster or the person's health status.

[0074] In the path planning process, the improved A* algorithm incorporates physical parameters into the heuristic function and risk avoidance mechanism. The heuristic function is modified to .

[0075] The Si of the weaker ones is lower, and the total cost of the path is The cost function g(n) increases significantly with the increase of the term, thus guiding the algorithm to prioritize flat, low-energy routes, even if the physical distance is slightly longer. The actual cost function g(n) combines the arrival time margin of the water front and dynamically balances time safety and physical feasibility through the risk aversion coefficient β. The specific formula is:

[0076] When Si is low, the system is synchronously improved and , give priority to assigning low-risk, low-physical exertion paths to weak people.

[0077] During the fine-grained correction phase, the system replans routes every three seconds based on the latest sensor data. For individuals with lower physical fitness, the algorithm avoids high-energy-demand areas like steep slopes and long alleyways, opting for detours. Meanwhile, individuals with higher physical fitness are allowed to choose higher-risk but more direct routes to shorten evacuation time. This differentiated strategy, through dynamic weight allocation and real-time data updates, achieves a coordinated optimization of risk and physical fitness, significantly improving the reliability of the emergency system and occupant survival rates.

[0078] Global layer (coarse-grained planning): Grid division: Divide the roadway network into coarse-grained grid cells of 5m×5m and establish a global topological map.

[0079] Algorithm: Based on the improved A* algorithm, the global shortest safe path is generated with the path length and the arrival time margin of the water front as the cost function.

[0080] Local layer (fine-grained correction): Dynamic subgrid division: The 5m grid of the global path is further subdivided into 1m×1m subgrids, increasing the resolution by 25 times.

[0081] Combined with real-time sensor data (water pressure, flow rate, personnel location), the improved A* algorithm is used to optimize local paths, adding physical attenuation factors and panic index corrections.

[0082] The distributed intelligent plugging device array adopts hydraulic drive and is equipped with a pressure-adaptive sealing mechanism, and its response delay time is controlled within 2 seconds.

[0083] Experiment 1: Emergency path planning and coordinated control of multi-node water inrush in underground coal mines.

[0084] Scenario description: A water inrush accident occurred in a fully mechanized coal mining face at a coal mine. The tunnel network, totaling 3.2 km in length, consisted of four main tunnels (4.5 m wide) and eight branch tunnels (3.2 m wide), with an average slope of 5°. A shearer accidentally penetrated an aquifer, triggering three water inrush points (with flow rates of 1.5 m³ / s, 0.8 m³ / s, and 2.2 m³ / s, respectively), creating a dynamic, spreading risk field. Twenty-three miners were located in the accident area, five of whom were in a high-risk area 500 m downstream of the water inrush point.

[0085] Multi-node water inrush emergency path planning and coordinated control in coal mines include the following steps: System deployment process: 1) Build an environmental monitoring network and deploy 32 multi-parameter sensors at tunnel intersections to collect real-time data on water pressure (accuracy ±0.1 MPa), flow (accuracy ±2%), and tunnel resistance coefficient. The data update frequency is 10 Hz. 2) Miners wear UWB positioning helmets to dynamically track their location information (positioning error ≤ 0.3m) and simultaneously monitor physiological parameters such as heart rate and movement speed; 3) Install 6 sets of hydraulic gate valves (response time 1.2s) at key nodes of the main tunnel to support remote coordinated blocking control.

[0086] Algorithm execution process: 1) Dynamic modeling and prediction of water inrush: Based on the water inrush diffusion prediction model, the spatiotemporal evolution of the water inrush front is predicted in combination with real-time sensor data. Calculations show that the gushing water will reach densely populated areas after 180 seconds, and the prediction error is reduced by 34% compared with traditional methods.

[0087] 2) Utilize the evacuation-blocking collaborative optimization model for path planning and improve the A* algorithm to generate two candidate paths: Path 1 (Traditional A*): Distance 480m, Estimated evacuation time 120s Path 2 (improved A* algorithm): distance 520m, but avoids high-resistance lanes, estimated evacuation time 98s.

[0088] 3) Collaborative optimization: Using a hybrid algorithm of genetic algorithm (GA) and particle swarm optimization (PSO) with a multi-objective function, the optimal blocking time tblock = 150s was obtained, with priority given to activating the three downstream gate valves.

[0089] Through the implementation of this technical solution, key performance indicators are significantly optimized:

[0090] This embodiment scheme verification: 1) Dynamic risk field response capability: Under extreme conditions where the speed of the water front suddenly changes by ±40%, the system still maintains a path update frequency of 1s, and the water diffusion prediction error is only 12%, which is 34% higher than the traditional model.

[0091] 2) The dynamic weight allocation algorithm increased the evacuation priority of seriously injured patients by 40%. Due to the dynamic weight adjustment mechanism, the actual evacuation time of seriously injured patients was shortened by 28% compared with the average, showing a significant effect of improving priority.

[0092] In Experiment 1, the effects of different β values ​​on personnel safety and energy consumption are verified as follows:

[0093] The optimal balance is achieved when β=0.3: the evacuation time is shortened by 28.6%, energy consumption is reduced by 39.7%, and the success rate of water inrush control reaches 92%.

[0094] Experiment 2: Collaborative control method of dynamic treatment and intelligent blocking of water inrush in subway tunnels.

[0095] Scenario Description: A sudden confined water layer penetration accident occurred during shield tunneling in an urban subway tunnel. The tunnel is 1.5 km long, and the lining structure includes five connecting tunnels (spaced 300 m apart). The water inrush pressure was 2.8 MPa, with an instantaneous flow rate of 3.5 m³ / s, resulting in a water accumulation rate of 1.2 m / min in the tunnel. At the time of the accident, 18 construction workers were located in the middle of the tunnel (800 m from the water inrush point) and in the rear area (1200 m from the water inrush point).

[0096] The method for coordinated control of dynamic treatment and intelligent blocking of water inrush in subway tunnels includes the following steps: Hardware system deployment process: 1) Layout of the environmental sensing layer: vibration-pressure composite sensors (sampling frequency 20 Hz) are deployed every 50 m along the tunnel axis to monitor water pressure fluctuations (accuracy ±0.05 MPa) and structural vibration anomalies in real time; 2) Personnel positioning layer arrangement: Construction workers wear LoRa + inertial navigation fusion positioning terminals (positioning error ≤ 1m) to simultaneously collect behavioral parameters such as movement speed and heart rate variation coefficient; 3) Execution control layer layout, deploy 4 sets of rapid expansion airbags (maximum pressure 3.5MPa, inflation time ≤5s) at key nodes of communication channels, supporting remote hierarchical triggering. System deployment Algorithm execution process: 1) Dynamic risk field modeling: Based on coupled analysis of fluid mechanics and structural mechanics, a three-dimensional water inflow diffusion model was constructed. Using real-time sensor data, the water front velocity was corrected to 1.2 m / s, improving prediction accuracy by 22% compared to traditional empirical models.

[0097] 2) Multimodal path planning, using an improved A* algorithm with a hierarchical optimization strategy to perform evacuation planning: Global layer: discretize the tunnel into 5m×5m grids to generate the initial evacuation path; Local layer: Dynamically adjust safety margins based on real-time panic index (movement speed drops by 33%) to ensure path feasibility; 3) Phased blocking decision-making, based on a hybrid algorithm of genetic algorithm (GA) + particle swarm optimization (PSO), to formulate an airbag collaborative blocking strategy: Phase 1 (t=60s): Activate the two downstream communication channel airbags to form the first water barrier; The second stage (t=90s): Start the upstream key node airbags and build a zoned drainage system.

[0098] Through the implementation of this technical solution, key performance indicators are significantly optimized:

[0099] This implementation plan verifies: 1) The hierarchical path planning architecture reduces global path calculation time from 15s to 3.2s, meeting sub-second dynamic response requirements; 2) Adaptive risk control: The dynamic adjustment mechanism of the β coefficient reduced the exposure time of people in high-risk areas by 41% without extending the total evacuation time; 3) Behavior perception is optimized, and the panic attenuation factor effectively offsets the impact of speed reduction. The actual path safety pass rate reaches 100%, an improvement of 27% compared to static planning.

[0100] Experiment 3: Coordinated treatment and intelligent hierarchical control method of multi-source water inrush in the diversion tunnel.

[0101] Scenario Description: A sudden multi-source water inrush incident occurred in a deep-buried hydraulic tunnel complex. The tunnels are 5 km long, comprised of three parallel branch tunnels (6m x 8m cross-section), and are buried at a maximum depth of 800 m. A geological survey error triggered simultaneous water inrush from branch tunnels 1 and 2 (total flow rate of 5.0 m³ / s, pressure gradient of 0.8 MPa / m), creating a highly dynamic seepage-stress coupling risk field. Thirty-five construction workers were located in the accident area, eight of whom were located in the high-pressure zone (pressure ≥ 4 MPa) in the branch tunnels.

[0102] The method for coordinated treatment and intelligent hierarchical control of water inrush from multiple sources in a diversion tunnel includes the following steps: Hardware system deployment process: 1) Distributed fiber optic sensor arrays (temperature monitoring accuracy 0.1°C, strain monitoring accuracy 10με) are deployed along the main tunnel and branch tunnels to capture seepage paths and structural deformation in real time; 2) Beidou differential positioning base stations are deployed outside the cave, and UWB positioning tags are used inside the cave (comprehensive positioning error ≤ 0.5m); 3) Two sets of high-pressure rapid gates (pressure bearing capacity 6MPa, locking time ≤3s) are installed at key nodes of the main tunnel to support hierarchical linkage control.

[0103] Algorithm implementation process: 1) The water inrush model, which considers both seepage and stress-strain coupling, predicts the arrival time of the main tunnel water front at 240 seconds (with an error of ±15 seconds). By correcting the seepage coefficient matrix using real-time fiber optic data, the confidence level in the water inrush evolution prediction was increased to 92%.

[0104] 2) Generation of hierarchical evacuation strategies: Using a multi-objective optimization algorithm driven by reinforcement learning to execute group evacuation planning, health levels (I-III) are divided according to vital signs data, with priority weights varying by 2.5 times: Group 1 (12 people, health level I): Evacuate in a straight line along the main tunnel (1.2 km), using rapid gate timing control to create a safety window; The second group (23 people, including 5 people with Class III health): detour via Branch Tunnel 3 (1.8 km), avoiding high-voltage areas and matching the mobility of personnel; 3) Multi-stage plugging optimization: Based on a hybrid algorithm of genetic algorithm (GA) and particle swarm optimization (PSO), the gate control strategy was optimized through 200 simulation iterations: Phase 1 (t=180s): Close the gate of branch tunnel 1, reducing the water flow by 45%; The second stage (t=210s): Close the gate of branch tunnel 2 and start the main tunnel drainage system simultaneously.

[0105] Through the implementation of this technical solution, key performance indicators are significantly optimized:

[0106] This embodiment scheme verification: 1) Intelligent hierarchical control: The health status grading mechanism increases the safety factor of the evacuation path for Class III health personnel by 60%, and no secondary injuries occur; 2) Reinforcement learning optimization: The model reduced blocking decision response time from 25 seconds to 6 seconds, improving strategy optimization efficiency by 316%; 3) Multi-field coupling modeling: The seepage-stress joint inversion algorithm reduces the water front prediction error from ±45s to ±15s, improving the accuracy by 233%.

[0107] Experiment 4: Intelligent prevention and control of water inrush disasters in deep mines and emergency evacuation coordination system.

[0108] Scenario Description: A sudden water inrush accident occurred in a deep coal mine due to a hidden water-conducting structure. The mine has a vertical depth of 920 meters and consists of four mining levels (-200 meters, -450 meters, -700 meters, and -920 meters). The main tunnels are 23.5 kilometers long, with an average slope of 7 degrees and a hydrogeological complexity level of III. The accident occurred at the -700-meter horizontal haulage tunnel, revealing a hidden cave (initial water inflow of 8.5 m³ / s, water pressure of 4.2 MPa), resulting in an instantaneous flooding rate of 1.8 m / min in adjacent tunnels. Sixty-seven workers were located within the accident area, 14 of whom were located in a low-lying area 300 meters downstream of the water inrush point (flooding depth ≥ 2 meters).

[0109] The deep mine water inrush disaster intelligent prevention and control and emergency evacuation collaborative system includes the following steps: Hardware system deployment process: 1) 48 sets of mine-grade intrinsically safe water inrush monitors (flow accuracy ±1.5%, water pressure monitoring range 0-10 MPa, sampling frequency 100 Hz) were deployed along the tunnels, and 16 laser water level scanners (resolution ±2 cm) were deployed at key nodes to construct a three-dimensional flooding field in real time.

[0110] 2) Miners are equipped with waterproof UWB precise positioning mining lamps (positioning error ≤ 0.4m), integrated body temperature and heart rate monitoring functions, and a personnel health status classification database (including physical index, chronic disease history, etc.) is established.

[0111] 3) Twelve sets of high-pressure waterproof and sealed doors (pressure resistance level 8MPa, locking time ≤3.5s) are arranged in the main tunnel, and eight high-power submersible pump sets are installed (single-unit drainage capacity 15m³ / min, supporting variable frequency linkage).

[0112] Algorithm implementation process: 1) Dynamic modeling of water inrush, integrated with real-time monitoring data, predicted the advance speed of the flooded front in the main transport tunnel to be 2.1 m / s (error ≤ 9%). Acoustic emission monitoring was used to invert the expansion of rock mass fractures, with a prediction accuracy of 88%; 2) Multimodal path planning: using an improved hierarchical A* algorithm to integrate the dynamic risk field to generate two optimized paths: At the global level, the roadway network is discretized into 5m×5m grids to generate the initial evacuation path; Local layer: Embed dynamic correction factors.

[0113] Route A: shortest distance 1.1 km (traversing the medium-risk area, suitable for healthy people); Route B: 1.6 km (detour via the elevated wind tunnel for people with reduced physical strength); 3) Drainage-plugging collaborative decision-making and resource scheduling optimization based on mixed integer programming: Level 1 response (t=90s): Activate the downstream 3# and 5# airtight doors to form an isolation tank; Secondary response (t=150s): Start pump groups 2# and 4# to establish a drainage buffer zone; Level 3 response (t=210s): Inject nano-quick-setting material (curing time ≤30s) to seal the cracks.

[0114] Through the implementation of this technical solution, key performance indicators are significantly optimized:

[0115] This embodiment scheme verification: 1) Dynamic risk field modeling: The seepage-stress coupling algorithm reduces the water front prediction error from ±25s to ±8s; 2) Hierarchical route optimization: The physical attenuation factor increased the evacuation efficiency of weak personnel by 52%, and no physical overdraft incidents occurred.

[0116] Through the implementation of this technical solution, event processing is significantly optimized. The comparative summary of the embodiments is as follows:

[0117] Calibrate the distributed optical fiber water pressure and water level monitoring system in the present invention: Equipment model: DPS-8000 distributed fiber optic water pressure and water level monitoring system.

[0118] Monitoring parameters: ±0.1% FS (full-scale water pressure measurement accuracy), 0.01 m water level resolution, 0-100 kPa (water pressure range) / 0-10 m (water level range) Standard equipment: Second-class standard piston pressure gauge (accuracy ±0.05% FS), high-precision laser level gauge (resolution ±0.005 m).

[0119] Calibration steps: Water pressure calibration: Connect the optical fiber sensor and the standard pressure gauge in series to a closed pressure vessel.

[0120] Set pressure points: 20 kPa, 40 kPa, 60 kPa, 80 kPa, 100 kPa.

[0121] Each pressure point was stabilized for 30 minutes and 5 measurement data were recorded.

[0122] Water level calibration: Fix the fiber optic sensor and the standard liquid level gauge vertically in the water tank.

[0123] Set the water level points: 1 m, 3 m, 5 m, 7 m, 9 m.

[0124] Each water level was stabilized for 20 minutes and 5 measurement data were recorded.

[0125] Water pressure calibration data results:

[0126] Water level calibration data results:

[0127] The maximum absolute error of water pressure measurement is +0.3kPa, meeting the nominal accuracy of ±0.1%FS (±0.1kPa). The maximum absolute error of water level measurement is +0.05 m, meeting the design requirement of 0.01 m resolution. The repeatability error is less than 0.2% of the range, and the system stability is good.

[0128] In Experiment 1, field data collection for multi-node water inrush emergency path planning and coordinated control in coal mines was conducted, and the following data were obtained:

[0129] Evacuation path: (100,200) → (150,200) → (200,180) → (300,180) → (500,220) → (800,250).

[0130] Path length: 720m Evacuation time: 132 seconds like Figure 2 As shown, the pressure drops sharply after the gate is activated.

[0131] This paper proposes a water diffusion prediction model based on multi-physics field coupling, improving prediction accuracy by over 30%. By integrating the three-dimensional spatial topology of tunnels, fluid dynamics, and porous media seepage theory, this model simplifies the Navier-Stokes equations into a water front propulsion model suitable for complex tunnel networks. Specifically, combining the Forchheimer seepage formula with dimensional analysis methods, a water front arrival time calculation model was established. This model maintains a prediction error within ±15% in simulations of similar materials (compared to ±40% for traditional models), significantly improving accuracy.

[0132] This invention establishes a collaborative optimization framework for evacuation paths and blocking strategies, achieving a 40% improvement in resource scheduling efficiency. This framework integrates personnel safe evacuation and blocking device energy consumption into a unified decision-making system through a multi-objective optimization model, using the blocking device startup time tblock as the decision variable to establish constraints. Using a hybrid strategy of genetic algorithm (GA) and particle swarm optimization (PSO) (GA population size 50, PSO inertia weight linearly decreasing from 0.9 to 0.4), the system improves convergence speed by 60% in a 20-person evacuation scenario and reduces blocking energy consumption by 39.7% (from 6.8×10 4 kJ dropped to 4.1×10 4 kJ). At the same time, the dynamic balancing logic adaptively adjusts the β value based on the scale of the water inflow: β is increased to reduce energy consumption during small-scale water inflow (flow rate <0.5m³ / s), and is reduced during large-scale water inflow (flow rate ≥2m³ / s) to prioritize personnel safety, ensuring that the system maintains a 92% water inflow control success rate even under extreme operating conditions.

[0133] The present invention develops a dynamic weight allocation algorithm that dynamically adjusts evacuation priorities based on personnel health status. This algorithm improves the A* path planning mechanism, introduces the physical fitness index (Si) to quantify personnel's real-time physiological state, dynamically modifies a heuristic function based on Si, and incorporates a risk avoidance mechanism for the arrival time margin of the water front. This algorithm increases the evacuation priority of seriously injured personnel by 40% and shortens actual evacuation time by 28% compared to the average. Furthermore, a physical fitness attenuation factor optimizes path selection for frail personnel, increasing evacuation efficiency by 52% in a deep mine case study, eliminating physical exhaustion incidents and significantly improving overall emergency response effectiveness.

[0134] Example 2 The purpose of this embodiment is to provide an emergency decision-making system for multi-node water inrush disasters in underground tunnels, including: A construction module is configured to: construct a three-dimensional dynamic model based on the roadway BIM model, roadway water inrush parameters, and personnel status; A prediction module is configured to: establish a water front arrival time prediction model based on the three-dimensional dynamic model and taking into account water inrush pressure, tunnel resistance, and fluid characteristics; The decision-making module is configured to: construct an evacuation-blocking collaborative optimization model based on personnel safety and the energy consumption of the blocking device, use an improved A* algorithm to generate a safe evacuation path for personnel, and obtain a safe evacuation plan and an optimal blocking plan with maximizing personnel safety and minimizing the energy consumption of the blocking device as constraints; wherein the improved A* algorithm introduces the personnel physical index to dynamically adjust the evacuation priority; and add the arrival time margin of the water front to the path cost function.

[0135] In further embodiments, there is also provided: An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed by the processor. When the computer instructions are executed by the processor, the method described in Example 1 is performed. For the sake of brevity, no further details are given here.

[0136] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), off-the-shelf field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0137] The memory may include a read-only memory and a random access memory, and provides instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0138] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in embodiment 1 is performed.

[0139] The method in Example 1 can be directly implemented as being executed by a hardware processor, or by a combination of hardware and software modules within the processor. The software module can be located in a storage medium well-established in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not given here.

[0140] A computer program product includes a computer program, and when the computer program is executed by a processor, the method described in embodiment 1 is implemented.

[0141] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions contained in program modules, which are executed in a device on a real or virtual processor of a target to perform the process / method described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided between program modules as needed. The machine-executable instructions for the program modules can be executed in local or distributed devices. In distributed devices, program modules can be located in local and remote storage media.

[0142] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the computer or other programmable data processing device, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0143] In the context of the present invention, computer program code or related data can be carried by any appropriate carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, and the like.

[0144] Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0145] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A method for emergency decision-making for multi-node water inrush disasters in underground tunnels, characterized in that: include: Based on the tunnel BIM model, tunnel water inrush parameters and personnel status, a water inrush diffusion prediction model is constructed. The diffusion path of water inrush in different branch tunnels and the time when the water inrush front reaches each tunnel node are predicted based on the water inrush diffusion prediction model. An evacuation-plugging collaborative optimization model was constructed based on personnel safety and the energy consumption of the plugging device. An improved A* algorithm was used to generate a safe evacuation path for personnel based on the water inrush prediction results. With maximizing personnel safety and minimizing the energy consumption of the plugging device as constraints, a safe evacuation plan for personnel and an optimal plugging plan were obtained. The improved A* algorithm introduced the personnel physical strength index to dynamically adjust the evacuation priority, and the arrival time margin of the water inrush front was added to the path cost function.

2. The method for emergency decision-making for multi-node water inrush disaster in underground tunnels according to claim 1, characterized in that: The time for the water front to arrive at each tunnel node is predicted, specifically: Taking into account the water gushing pressure, tunnel resistance and fluid characteristics, a water gushing front arrival time prediction model was established. The time when the water gushing front arrived at each tunnel node was predicted based on the water gushing front arrival time prediction model.

3. The method for emergency decision-making for multi-node water inrush disaster in underground tunnels according to claim 1, characterized in that: The heuristic function of the improved A* algorithm for: ， ， ， Among them, f(n) is the estimated total cost of node n; g(n) is the actual cost from the starting point to node n, is the physical sensitivity coefficient, The real-time moving speed of personnel, is the resting heart rate, is the personnel’s current heart rate, is the risk aversion coefficient, is the personnel physical strength index, is the individual's maximum heart rate, is the maximum movement speed of the individual, is the resting heart rate.

4. The method for emergency decision-making for multi-node water inrush disaster in underground tunnels according to claim 1, characterized in that: Based on the water inrush prediction results, a safe evacuation path for personnel is generated. With the constraints of maximizing personnel safety and minimizing the energy consumption of the plugging device, a safe evacuation plan for personnel and an optimal plugging plan are obtained. Specifically, Taking the plugging device activation time as the decision variable, a personnel safety function is constructed based on the distance between personnel and the water front. Taking the plugging device start-up time as the decision variable, the plugging device energy consumption function is constructed according to the plugging device's hierarchical energy consumption coefficient. Based on the constructed personnel safety function and the blocking device energy consumption function as an evacuation-blocking collaborative optimization model, it is determined whether to quickly activate the blocking device according to the evacuation-blocking collaborative optimization model; When it is determined that the plugging device needs to be quickly activated, the activation priority and time window of the plugging device are dynamically adjusted through the reinforcement learning algorithm to generate the optimal plugging plan.

5. The method for emergency decision-making for multi-node water inrush disaster in underground tunnels according to claim 4, characterized in that: When it is determined that the plugging device needs to be quickly activated, the activation priority and time window of the plugging device are dynamically adjusted through the reinforcement learning algorithm to generate the optimal plugging solution, specifically: The parameters of the start time of each plugging device, the start queue order, the interval time between adjacent plugging devices and the energy consumption coefficient are combined into a plugging plan; Genetic algorithms are used to perform crossover and mutation operations on the scheduling scheme to obtain a globally filtered blocking scheme; The particle swarm optimization algorithm is used to screen the global optimization scheme to obtain the locally optimized plugging scheme; The optimal plugging solution is obtained based on the quantitative score of the locally optimized plugging solution.

6. The method for emergency decision-making for multi-node water inrush disaster in underground tunnels according to claim 5, characterized in that: A multi-objective fitness function is constructed to quantify the score of the locally optimized plugging solution. The multi-objective fitness function is composed of the ratio of the actual number of people who can be safely evacuated to the total number of people who need to be evacuated, the total energy consumption of the plugging device, and the ratio of the time window margin to the maximum allowable time margin. The time window margin is determined by the difference between the plugging start time window and the arrival time of the water front.

7. An emergency decision-making system for multi-node water inrush disasters in underground tunnels, characterized by: include: The prediction module is configured to: construct a water inrush diffusion prediction model based on the roadway BIM model, roadway water inrush parameters, and personnel status; and predict the diffusion path of water inrush in different branch roadways and the time when the water inrush front arrives at each roadway node based on the water inrush diffusion prediction model; The decision-making module is configured to: construct an evacuation-blocking collaborative optimization model based on personnel safety and the energy consumption of the blocking device; use the improved A* algorithm to generate a safe evacuation path for personnel based on the water inrush prediction results; and use maximizing personnel safety and minimizing the energy consumption of the blocking device as constraints to obtain a safe evacuation plan for personnel and an optimal blocking plan; among them, the improved A* algorithm introduces the personnel physical index to dynamically adjust the evacuation priority; and adds the arrival time margin of the water inrush front to the path cost function.

8. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 6 is completed.

9. A computer-readable storage medium, characterized in that Used to store computer instructions, which, when executed by a processor, complete the method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The invention comprises a computer program, which is used to implement the method according to any one of claims 1 to 6 when the computer program is executed by a processor.

Images