Chemical industry park emergency treatment method based on digital twinning

Abstract

A chemical industry park emergency processing method based on digital twinning carries out high-precision digital modeling on the entity environment of the chemical industry park, and realizes real-time sensing and reflection of dynamic changes in the park; the Internet of Things device model is fused with the twin scene, the Internet of Things device model is matched with the actual park enterprise device information, and the model and the monitoring device are cooperatively linked; a variety of chemical industry park emergency scene simulation is provided, different scenes are selected for drilling according to actual needs; a chemical industry park accident emergency resource scheduling model is constructed, digital twinning simulation technology and a hybrid intelligent optimization algorithm are fused, and resource allocation under dynamic and complex environment is realized; the necessary point and the bypass area are comprehensively considered, and a collision-free safe path from the starting point to the ending point is found. The present application provides a comprehensive emergency simulation drilling system and method for the chemical industry park by integrating high-precision digital twinning technology, so as to comprehensively optimize and improve the emergency response capability and safety management level of the park.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of safety management, and particularly relates to a chemical industrial park emergency processing method based on digital twinning. BACKGROUND

[0002] In the field of chemical industrial park safety management, traditional emergency simulation drills usually rely on physical simulation, paper-based deduction or simple computer simulation. Physical simulation sets up a simulation scene in the chemical industrial park or a specific site, organizes personnel to conduct field drills, which is costly and is easily restricted by site, weather and other conditions; paper-based deduction draws emergency flowcharts, formulates emergency response plans, and deduces on paper, which is less costly but lacks intuitiveness and interactivity, making it difficult to comprehensively evaluate the actual effect of emergency response; computer simulation simulates emergency events in the chemical industrial park by using computer software, which has high flexibility and scalability, but is limited by the development level of computer technology, and the simulation accuracy and fidelity are often insufficient.

[0003] Traditional chemical industrial park emergency simulation drill schemes have deficiencies in accuracy, cost, real-time performance and evaluation optimization, SUMMARY

[0004] The core purpose of the present application is to provide a comprehensive emergency simulation drill system and method for a chemical industrial park by integrating high-precision digital twinning technology, so as to comprehensively optimize and improve the emergency response capability and safety management level of the park.

[0005] The chemical industrial park emergency processing method based on digital twinning includes the following steps:

[0006] Step S1: Constructing a high-precision digital twinning model: high-precision digital modeling of the entity environment of the chemical industrial park is performed to accurately restore the equipment, facilities and buildings in the park, and to real-time perceive and reflect the dynamic changes in the park;

[0007] Step S2: Integrating Internet of Things device models with twinning scenarios, matching the Internet of Things device models with the actual park enterprise device information to achieve coordinated linkage of the models and monitoring devices;

[0008] Step S3: Providing multiple chemical industrial park emergency scenario simulations, including at least fire, explosion, chemical leakage, and selecting different scenarios for drills according to actual needs;

[0009] Step S4: Constructing a chemical industrial park accident emergency resource scheduling model, integrating digital twinning simulation technology and hybrid intelligent optimization algorithm to achieve efficient allocation of resources (such as fire fighting equipment, medical rescue supplies, protective equipment, etc.) in dynamic and complex environments, significantly improving scheduling response speed, resource utilization and accident response efficiency.

[0010] Step S5: Considering the mandatory points and the detour area, a collision-free safe path from the starting point to the ending point is found within the specified range of the target object in the park.

[0011] Compared with the prior art, the present application has the following advantages:

[0012] Improve the realism and accuracy of emergency simulation: By constructing a digital twin model that accurately maps and highly matches the real chemical industry park, the precise simulation of various emergency events is realized, thereby improving the realism and accuracy of simulation drills, and making the drill effect more close to the actual accident scene.

[0013] Reduce the cost of emergency simulation drills: Relying on digital twin technology for simulation drills, without investing a large amount of site, equipment and manpower cost, the simulation scene constructed and the data accumulated can be repeatedly called and deeply analyzed, avoiding the repeated waste of resources, while ensuring the quality and effect of the drill, realizing the substantial reduction of the drill cost

[0014] Enhance the real-time and effectiveness of emergency response: Through the real-time data acquisition and transmission system, real-time monitoring and response to emergency events are realized, the initial signs, development dynamics and key change nodes of the event are quickly captured, and the speed and efficiency of emergency response are improved.

[0015] Establish a scientific emergency evaluation system: Use data mining and machine learning algorithms to deeply analyze and mine simulation drill data, accurately extract key information and potential laws contained therein, and then establish a scientific evaluation standard and index system to provide strong support for the optimization and improvement of emergency response plans. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The present application DETAILED DESCRIPTION

[0017] The technical solutions of the present application will be described in detail below in conjunction with the drawings:

[0018] As Figure 1 shown, the steps of the chemical industry park emergency handling method based on digital twinning are as follows:

[0019] First, build a high-precision digital twin model: high-precision digital modeling of the physical environment of the chemical industry park is performed to accurately restore equipment, facilities, buildings, etc. within the park, and to real-time perceive and reflect dynamic changes within the park, such as equipment operating status, personnel flow, etc.

[0020] Second, fuse the Internet of Things device model with the twin scene, match the Internet of Things device model with the actual park enterprise device information, and realize the coordinated linkage of the model and the monitoring device.

[0021] Third, provide a variety of chemical industry park emergency scene simulation, including fire, explosion, chemical leakage, etc., can be selected according to the actual needs of different scenes for training.

[0022] Fourth, build a chemical industry park accident emergency resource scheduling model, integrate digital twin simulation technology and hybrid intelligent optimization algorithm, realize efficient resource allocation (such as fire fighting equipment, medical rescue materials, protective equipment, etc.) in dynamic and complex environment, significantly improve the scheduling response speed, resource utilization and accident response efficiency.

[0023] Fifth, considering the necessary points and bypass areas, the target object finds a collision-free safe path from the starting point to the ending point within the specified range of the park.

[0024] Sixth, after the exercise, the exercise process is evaluated, and an evaluation report is generated, including exercise time, task completion, personnel performance, etc. Users can summarize lessons learned from the evaluation report and improve emergency plans.

[0025] Chemical industry park emergency simulation and drilling in the process of function execution have post execution level personnel, level command personnel, expert team, guide control personnel participate together, according to a standard process from the occurrence to the disposal end of an accident Control complete the three-dimensional digital twin simulation exercise process, while tracking the whole exercise process, so as to realize the examination and evaluation of the simulation exercise process.

[0026] The technical solutions of the present application will be described in detail as follows:

[0027] 1. High-precision digital twin chemical industry park scene construction

[0028] Digital twin visualization engine: use efficient and stable 3D rendering engine to provide foundation for chemical industry park twin scene construction and visualization expression, including multi-dimensional space data visualization, rendering mechanism, environment simulation, video scene fusion, etc. Based on the powerful rendering capability of cloud rendering engine server, relying on network information flow, it provides fine scene display for each terminal while eliminating the difference of front-end hardware, realizes fast and efficient loading and browsing of scene, and supports interactive operation of users.

[0029] 2. Internet of Things device data fusion processing

[0030] Internet of Things device model and twin scene fusion: match the Internet of Things device model with the actual park enterprise device information, realize the cooperation of model and monitoring device, collect the key position, video, electricity, sound, heat, light and other data of monitoring object in real time, and access the existing related monitoring, monitoring, model data, and realize the dynamic collection of key data of sensing object.

[0031] 3. Accurate simulation of emergency scenarios

[0032] Allow the drill organizer to set up diverse emergency drill scenarios based on the digital twin model according to different drill objectives and needs. Different accident types (such as fire, explosion, leakage, etc.), accident locations (specific enterprises or areas within the park), accident scales (small-scale local accidents or large-scale park-level accidents), and initial environmental conditions (such as weather conditions, surrounding traffic conditions, etc.) can be selected. Advanced physical engines and simulation algorithms are used to simulate the development process of emergency events. The algorithm should be able to accurately simulate key processes such as fire spread, toxic gas diffusion, and personnel evacuation, and consider the influence of environmental factors such as wind speed, temperature, and humidity on the simulation results.

[0033] (1) Construct a hazardous substance leakage and diffusion scenario in a chemical industrial park. Based on the principles of fluid mechanics and mass transfer, construct a diffusion model of hazardous substances in the atmosphere, water body, and soil.

[0034] (2) Construct a fire spread scenario in a chemical industrial park. Use a model based on the principles of energy conservation and heat transfer to simulate the development process of the fire.

[0035] (3) Construct an explosion accident scenario in a chemical industrial park. For explosion accidents, construct a comprehensive model including explosion shock wave propagation, explosion debris scattering, and secondary disasters (such as fire, toxic substance diffusion, etc.) caused by explosion.

[0036] 4. Emergency resource scheduling

[0037] Construct an emergency resource scheduling model for a chemical industrial park, integrate digital twin simulation technology and hybrid intelligent optimization algorithm, and realize efficient allocation of resources (such as fire fighting equipment, medical rescue materials, protective equipment, etc.) in dynamic and complex environments, significantly improve the response speed, resource utilization rate and accident response efficiency.

[0038] 4.1 Interaction mechanism between agent and digital twin scenario

[0039] (1) Resource supply agent

[0040] Obtain accurate resource reserve information of resource supply points in real time through digital twin scenario, including detailed data such as the number of each resource, storage location, expiration date, etc. At the same time, according to the prediction of resource consumption trend by digital twin model, combined with reinforcement learning algorithm to decide resource allocation plan.

[0041] Resource reserve information in digital twin scenario: including the current actual quantity of various resources (such as fire fighting equipment, protective equipment, medical materials, etc.) at each supply point Storage location coordinates, etc.

[0042] State space:

[0043]

[0044] Action space:

[0045]

[0046] Resource types allocated by each supply point and quantity

[0047] Resource replenishment time information i.e. the estimated time needed from the moment of resource replenishment request to the actual replenishment of resources.

[0048] Digital twin predicted resource consumption rate The speed of consumption of each resource per unit time predicted based on factors such as the type, scale, and development trend of the accident.

[0049] Reward function R:

[0050] 1) Rescue time reward function:

[0051]

[0052] where T windows is the optimal rescue time window of the accident simulated by the digital twin, (Ti is the completion time of the ith rescue sub-task);

[0053] 2) Resource utilization rate reward function:

[0054]

[0055] where is the digital twin predicted demand for the kth resource, and δ is the allowed error range.

[0056] Using reinforcement learning algorithm, the agent selects action A supply according to the current state S supply , through the policy network π θ (a|s)(network parameters), and updates learning according to the reward function R, the goal is to maximize the long-term cumulative reward (γ is the discount factor).

[0057] (2) Transportation agent

[0058] Based on the dynamic information of the road network in the digital twin scene, such as real-time traffic flow, road conditions (whether congested, damaged due to accidents, etc.), and the location and state information of transportation tools in the digital twin space, the optimal transportation route is planned and the transportation strategy is adjusted. The interaction of transportation agents with the digital twin scene also includes real-time simulation of the transportation process, prediction of potential problems that different transportation schemes may encounter, and early avoidance.

[0059] Road network information G = (V, E) in digital twin scene: including nodes V (resource supply points, accident sites, road intersections, etc.) and edges E (road connection relationships), as well as the basic attributes of each edge (such as length, width, etc.).

[0060] Real-time traffic conditions C ij : provided by Internet of Things sensing devices and digital twin scene fusion, including congestion levels (represented by congestion coefficient ρ ij , 0 ≤ ρ ij≤1 , 0 represents smooth, 1 represents complete congestion), speed v ij , etc. information of each road.

[0061] Location L t and state S t of transportation tools: real-time geographic location coordinates of transportation vehicles and running state (such as whether full load, whether failure, etc.) of the vehicles obtained through Internet of Things sensing devices.

[0062] Real-time weather conditions W condition in digital twin scene: such as wind speed ω speed , wind direction ω direction , rainfall intensity r internsity , etc., which will affect transportation safety and efficiency.

[0063] Potential interference A interferense of other related accidents in the surrounding area to the transportation route: for example, the range and expected duration of road closures or traffic controls caused by nearby accidents.

[0064] State space: S transport = {G, C ij , L t , S t , W condition , A interferense}.

[0065] Action space: A transport = {Route ij , Vehicle m , V adjust}.

[0066] Reward function:

[0067] Transportation cost reward function

[0068] where C transport =∑ i (C fuel-i +C depreciation-i +C labor-i ), C fuel-i = S ij ×c fuel ×(1+α ω ×ω speed +α r ×r intensity )(c fuel is the fuel consumption cost per unit distance, α ω and α r are the coefficients of the impact of wind speed and rainfall intensity on fuel consumption), ΔC transport is the cost increase due to dynamic changes in road conditions (such as temporary restrictions or repairs), C budget is the preset transportation cost budget.

[0069] Similarly, using reinforcement learning algorithm, action A transport is selected according to state S transport , and learning update is carried out according to the reward function, so as to optimize the transportation strategy, reduce the transportation cost and time, and ensure the transportation safety.

[0070] (3) Rescue scene agent

[0071] The digital twin scene is used to comprehensively perceive the detailed situation of the accident scene, such as the accurate type of the accident (not only the large categories of fire, leakage, etc., but also the specific chemical leakage categories, fire combustion characteristics, etc.), the real-time development situation of the accident (fire spread direction, leakage diffusion range and speed, etc.), and the use of existing resources on the scene. Through deep interaction with the digital twin scene, the rescue scene agent can more accurately assess resource demand and provide more targeted resource demand adjustment suggestions to other agents.

[0072] Accident information in the digital twin scene: including the type T accident of the accident (such as fire, explosion, leakage, etc.), real-time development situation information of the accident, such as fire spread direction and speed, diffusion range and speed of the leakage, etc.

[0073] Use of existing resources on the scene: types and quantities of resources that have arrived at the accident scene, and use progress of the resources (such as consumption proportion P foam-consume of fire-fighting foam, etc.).

[0074] Accident Development Trend Prediction T in Digital Twin Scenario Simulation accident-trend For example, the estimated duration T of the fire fire-predict Predicted maximum extent of leaked material diffusion (L) max-spread-predict wait.

[0075] Resource demand adjustment recommendations: Propose the types of resources that need to be allocated to the resource supply agent and the transportation agent. and quantity The proposed adjustments.

[0076] State space:

[0077] The severity of the accident, S severity The progress of the rescue operation can be quantitatively assessed based on indicators such as the scope of the accident's impact and the amount of hazardous material leaked. rescue This can be assessed by the percentage of completed rescue missions or key indicators (such as the fire suppression rate, P). fire-extinguish (etc.) indicates.

[0078] Action space:

[0079] Reward function:

[0080] Accident Loss Reward Function

[0081] Where L accident-predict For real-time assessment and prediction of accident losses in digital twin scenarios, including casualty prediction. casaultu-predict Property loss prediction L property-predict and environmental damage prediction L environment-predict L max The maximum acceptable accident loss threshold.

[0082] Based on the reinforcement learning algorithm, according to state S scene Generate action A scene It learns and optimizes based on the reward function to improve rescue efficiency and reduce accident losses.

[0083] 4.2 Network Flow Optimization Module Empowered by Digital Twins

[0084] (1) Network construction and dynamic parameter update

[0085] Resource supply point information in a digital twin scenario: including the location coordinates of the supply point and the maximum resource reserve. And the proportion of supply capacity reduction β due to surrounding accidents i (For example, if a supply point is unable to allocate some supplies due to a nearby fire, β) i >0).

[0086] Accident site information: accident location coordinates, resource demand dynamic fluctuation predicted by digital twin scene (Changes in resource demand as the accident develops, such as expansion of dangerous areas or increase in rescue difficulty).

[0087] Road network information: including nodes V'(resource supply points, accident sites and possible transit nodes) and edges E'(road connection relationships), as well as dynamic changes in road capacity monitored by digital twin scene (Such as damage to some sections due to surrounding building fires, reducing road capacity) and real-time adjustment of transportation time (Calculated according to real-time traffic conditions, weather conditions and road condition changes, for example, Where is the dynamically adjusted transportation speed.

[0088] Updated network flow model: including adjusted network node set V', edge set E', edge capacity matrix C'(where ), transportation time matrix T'(where ) and resource demand vector Q required (Where

[0089] Model formula and logical relationship:

[0090] Construct network N=(V',E',C',T',D')

[0091] Where D' is the dynamic parameter matrix driven by the digital twin scene, containing and and other information.

[0092] For example, the resource supply constraint is adjusted to To consider the case of supply point supply capacity decline; the capacity limit constraint is updated to To ensure that resource transportation flow meets the actual road capacity.

[0093] (2) Dynamic adjustment of optimization objectives and constraint conditions

[0094] Updated network flow model parameters (such as V', E', C', T', Q required and so on).

[0095] Degree of reduction in accident risk P evaluated by digital twin scene risk-reduce : The accident spread range reduction ratio α spread-reduce , dangerous substance concentration reduction ratio α concentration-reduce and other index quantitative evaluation by digital twin scene simulation, for example, R risk-reduce = ω1α spread-reduce + ω2αconcentration-reduce ω1 and ω2 are weight coefficients.

[0096] Final output of optimized resource scheduling scheme: determine the resource transportation flow on each transportation route Final allocation of resource supply points, etc. to meet the optimization objectives and constraints.

[0097] Model formula and logical relationship:

[0098] Objective function Z = λC total +(1-λ)T total +μR risk-reduce

[0099] Where, T is calculated as described in the transportation agent part, total =∑ (i,j)∈∈ T′ ij x ij .

[0100] Constraints:

[0101] 1) Resource supply constraints: (as described above considering the constraints after the supply capacity decreases).

[0102] 2) Resource demand constraints:

[0103] 3) Flow conservation constraints: for transfer nodes k∈V′ other than resource supply points and accident sites, i:(i,k)∈E′ x ik ≤∑ j:(k,j)∈E′ x kj In the case of new nodes or node failure in the digital twin scenario, dynamically adjust the calculation range and node relationship.

[0104] 4) Capacity constraints:

[0105] Adaptive network flow optimization algorithm is adopted to gradually approach the optimal solution by constantly finding augmented paths and adjusting network flow to meet the objective function and constraints.

[0106] 4.3 Deep coordination mechanism of digital twin, agent and network flow

[0107] (1) Real-time data sharing and synchronization

[0108] Real-time data in the digital twin scenario: including the updated status data of the accident site every second (such as the development of the accident, resource usage, etc.), real-time traffic data of the road network, and resource reserves and consumption data of the resource supply point every second.

[0109] Decision information of multi-agent: resource allocation decision of resource supply agent and Transport route selection Route of transport agent ij , transport vehicle selection Veh icle m and transport speed adjustment strategy V adjust Resource demand adjustment suggestion of rescue site agent and

[0110] Optimization result of network flow optimization module: optimized resource scheduling scheme, i.e. resource transportation flow on each transport route and other information.

[0111] Push the data in the digital twin scene to the multi-agent reinforcement learning module and the network flow optimization module respectively; deliver the multi-agent decision information to the digital twin scene for simulation evaluation, and synchronize the evaluation results to the network flow optimization module; feedback the optimization results of the network flow optimization module to the digital twin scene, and then deliver them to the multi-agent by the digital twin scene.

[0112] Establish data sharing interface and communication protocol to ensure real-time and accurate transmission of data between digital twin scene, multi-agent reinforcement learning module and network flow optimization module. For example, use message queue or WebSocket technology to realize asynchronous data transmission and subscription publishing mode, and ensure the timeliness and integrity of data.

[0113] (2) Collaborative decision-making and dynamic adjustment

[0114] Synchronization data and evaluation results of digital twin scene: such as simulation evaluation of multi-agent decision effect (such as resource arrival time caused by certain resource allocation decision, influence on accident development, etc.), simulation verification results of network flow optimization scheme in digital twin scene (such as whether there is a transport bottleneck, whether the resource allocation is reasonable, etc.).

[0115] Current state information of multi-agent: including resource reserve change state S supply of resource supply agent, transport process state S transport of transport agent, and latest state S scene of accident site of rescue site agent.

[0116] Optimization scheme information of network flow optimization module: such as resource transportation flow allocation resource supply point allocation plan, etc.

[0117] Output dynamic adjustment decision: multi-agent adjusts its own decision according to the feedback of digital twin scene and network flow optimization result, such as resource supply agent adjusts resource allocation amount and type and Transport agent adjusts transport route Route ij-new , transport speed V adjust-new , rescue site agent adjusts resource demand suggestion and Network flow optimization module further optimizes resource scheduling scheme according to the new decision of multi-agent and the dynamic change of digital twin scene, and generates new resource transport flow , etc.

[0118] In the multi-agent reinforcement learning module, each agent recalculates the action selection strategy according to the new input state and the feedback information obtained from the digital twin scene and the network flow optimization module. For example, the policy update formula of the resource supply agent can be expressed as:

[0119]

[0120] Where θ supply is the policy network parameter of the resource supply agent, α supply is the learning rate, is the policy function, is the new resource allocation action, is the new state, is the new reward (based on resource allocation effect evaluation and other factors in digital twin scene).

[0121] The network flow optimization module reconstructs the network flow model and performs optimization calculation according to the new decision of multi-agent and the dynamic parameter change in the digital twin scene (such as further deterioration or improvement of road conditions, sudden change of resource demand at the accident site, etc.). For example, when the transport agent changes the transport route due to the prompt of the digital twin scene that the traffic condition of a certain road section deteriorates sharply, the network flow optimization module adjusts the edge capacity, transport time and other parameters related to the route accordingly, and recalculates the objective function:

[0122]

[0123] Where are the total cost, total time and accident risk reduction degree calculated based on the new network flow model and the development of the accident, respectively, and the new resource transport flow is obtained by solving, so as to realize the dynamic optimization of resource scheduling scheme.

[0124] The whole process of collaborative decision-making and dynamic adjustment continues to circulate in the process of accident emergency handling until the accident is effectively controlled and handled. Each cycle is based on the latest data provided by the digital twin scene and the close cooperation between modules, continuously optimizing the emergency resource scheduling strategy to adapt to the complexity, uncertainty and dynamic changes of the chemical industry park accident, maximizing the efficiency of resource scheduling, reducing accident losses and ensuring the safety and stability of the park.

[0125] 5. Emergency rescue and evacuation path planning

[0126] In the process of emergency response in the chemical industry park, road conflicts or congestion may occur due to the simple road network and weak traffic capacity in the park, resulting in evacuation failure or rescue vehicles unable to evacuate. Rescue / evacuation path planning takes into account the necessary points and bypass areas to find a collision-free safe path from the starting point to the end point within the park's specified range.

[0127] 5.1 Dynamic grid method environment modeling

[0128] The dynamic grid method is used for environment modeling, which divides the environment of the chemical industry park into small square cells, i.e. grids. Each grid represents an area in the environment and can be used to represent obstacles, free space, etc. By updating the state of the grid, the environment map can be gradually constructed.

[0129] (1) Representation and initialization of grid

[0130] Each grid has its corresponding state attribute, which can be defined as three basic states: free (passable), obstacle (non-passable), and unknown (initially undetected state).

[0131] Let S(G ijk ) represent the state of grid G ijk , which can be initialized as follows:

[0132]

[0133] At the beginning of the chemical industry park environment modeling, for those obviously known road areas corresponding to the grid, S(G ijk ) = 0; for the areas occupied by buildings, large chemical equipment, etc. corresponding to the grid, S(G ijk ) = 1; and for those areas that have not been detected by sensors, S(G ijk ) = -1.

[0134] The three-dimensional grid map M of the whole chemical industry park can be represented as the set of all grid states, i.e.

[0135] M = {S(G ijk| i = 1, 2,..., I; j = 1, 2,..., J; k = 1, 2,..., K

[0136] where I, J, K are the total number of grids in the three dimensions of row, column and layer respectively.

[0137] (2) Raster state update based on sensor information

[0138] Assuming that the sensor detects the environment, the detection result has certain uncertainty (for example, the sensor has a certain false alarm rate and a false negative rate), and the Bayesian filter is used to update the probability estimation of the raster state.

[0139] Let P(S(G ijk ) = s | Z 1:t ) represent the probability of the grid G 1:t being in state s (s takes the value of 0, 1 or -1) after receiving a series of sensor observations Z ijk from time 1 to time t.

[0140] According to the Bayesian formula, the updating process is as follows:

[0141] 1) Prediction step (prior probability update):

[0142]

[0143] Here P(S(G ijk ) = s | S(G ijk ) = s', Z 1:t-1 ) represents the transition probability of the grid becoming state s at the current time under the condition of knowing that the previous grid is in state s' and the previous observation, which depends on the dynamic characteristics of the environment and the performance of the sensor and other factors;

[0144] P(S(G ijk ) = s' | Z 1:t-1 ) is the probability of the previous grid being in state s', which comes from the update result of the last round.

[0145] 2) Update step (posterior probability update combined with current observation):

[0146]

[0147] where P(Z t | S(G ijk ) = s) represents the probability of obtaining the observation Z t at time t when the grid G ijk is in state s, which is determined by the observation model of the sensor.

[0148] For example, if the sensor detects that a certain grid area which was originally unknown (S(G ijk ) = -1) has an obstacle (observation indicates obstacle), through the calculation process of the above Bayesian filter, the probability of updating the grid state to obstacle (S(G ijk ) = -1) can be obtained, and when the probability exceeds a certain threshold (such as 0.8), it can be determined that the state is an obstacle.

[0149] (3) Grid map construction and visualization

[0150] In order to visualize the grid map in three-dimensional space, it is necessary to convert the row and column coordinates of the grid into actual three-dimensional space coordinates. Let the edge length of the grid in the x, y, z directions be d x , d y , d z (unit: meter or other length unit), and the lower left corner coordinates of the grid G ijk be (x ijk , y ijk , z ijk ), then:

[0151] x ijk = j × d x

[0152] y ijk = i × d y

[0153] z ijk = (k-1) × d z

[0154] In order to more accurately perform graphical rendering and other operations, the center coordinates of the grid need to be obtained, and let the center coordinates of the grid G ijk be (x ijk , y ijk , z ijk ), and the calculation formula is as follows:

[0155]

[0156] 5.2 Mathematical modeling of bidirectional path planning

[0157] Emergency response is usually divided into two stages: intra-campus emergency response and inter-campus collaborative emergency response. The goal is to plan the optimal bidirectional path for emergency rescue and emergency evacuation in different stages, respectively, to achieve the efficiency of rescue and evacuation, while avoiding road conflicts and congestion.

[0158] The starting and ending points of emergency rescue and evacuation, the locations of nodes along the route, the travel time of each road segment, and the flow of rescue and evacuation are set as decision variables. By optimizing these variables, the optimal route and resource allocation scheme are determined.

[0159] These include road capacity limits, i.e. the maximum traffic or pedestrian flow that each road segment can accommodate; time limits, such as completing rescue and evacuation within the specified emergency response time; and node flow balance constraints, ensuring that the flow entering and leaving each node is equal to guarantee the continuity and rationality of the path.

[0160] (1) Basic elements and variable definitions of the model

[0161] 1) Spatial Representation and Raster Definition

[0162] Assuming the three-dimensional space of the chemical industrial park is discretized into a series of cubic grids, denoted by G... ijk Let i represent the grid cell located in the i-th row, j-th column, and k-th layer (i, j, and k are all positive integers).

[0163] 2) Setting decision variables

[0164] Introducing decision variable x ijk This is used to represent path planning status. If a path passes through grid G... ijk Then x ijk =1; if the path does not pass through this grid, then x ijk =0.

[0165] 3) Definition of starting point and ending point

[0166] Emergency Rescue: Let the coordinates of the emergency rescue starting point be (i r ,j r ,k r The corresponding rescue target endpoint coordinates are: For example, the starting point of a rescue operation might be the grid location of the emergency rescue command center within the park, while the endpoint is the grid corresponding to key rescue points such as the core area where the accident occurred.

[0167] Emergency evacuation: The starting point coordinates for emergency evacuation are set to (i e ,j e ,k e The coordinates of the evacuation target endpoint are: For example, the starting point of the evacuation is the grid position in the densely populated area within the park, and the ending point is the grid corresponding to the safe assembly area outside the park.

[0168] 4) Definition of other relevant parameters

[0169] Distance metric: using Indicates from grid To grid the distance.

[0170] Travel Time: denotes the time required for traveling from grid to grid , which is affected by various factors such as road conditions within the grid, the speed of personnel or vehicles, the presence of dangerous situations, etc.

[0171] Travel Cost: is used to measure the comprehensive travel cost from grid to grid , taking into account factors such as distance, concentration of dangerous substances, and road congestion to determine the specific value.

[0172] (2) Constraints

[0173] 1) Path Connectivity Constraint

[0174] In three-dimensional space, the grids on the path need to maintain connectivity, that is, for any two consecutive grids and they should be adjacent in space. For a three-dimensional grid, a grid usually has 26 neighboring grids (including its own plane and the neighboring planes above and below).

[0175] Define the neighborhood function N(G i,j,k ) to represent the set of all neighboring grids of grid G i,j,k , then the mathematical formula of the path connectivity constraint is:

[0176]

[0177] The meaning of this formula is that if a certain grid is on the planned path (i.e. ), then at least one of its neighboring grids must also be on the path (i.e., the sum of for its neighboring grids is at least 1), to ensure that the planned path is continuous and passable.

[0178] 2) Start and End Point Constraints

[0179] For emergency rescue paths and emergency evacuation paths, there are clear requirements for the start and end points, and the corresponding constraint formulas are as follows:

[0180] Emergency Rescue Path Start Point Constraint:

[0181]

[0182] The rescue path must start from the set rescue start point, so the decision variable x corresponding to the start grid is assigned a value of 1.

[0183] Emergency rescue path end point constraint:

[0184]

[0185] Ensure that the rescue path can eventually reach the set rescue target end point, so the decision variable x corresponding to the end grid is also assigned a value of 1.

[0186] For emergency evacuation paths, there are similar constraints:

[0187] Emergency evacuation path start point constraint:

[0188]

[0189] Emergency evacuation path end point constraint:

[0190]

[0191] 3) Flow conservation constraint (considering multiple agents traveling at the same time)

[0192] When there are multiple rescue teams or multiple batches of evacuees simultaneously planning paths, in order to ensure flow balance, the flow entering a grid should be equal to the flow leaving the grid.

[0193] Let f in (i,j,k) represent the flow into the grid, and f out (i,j,k) represent the flow out of the grid G ijk , then the mathematical expression of the flow conservation constraint is:

[0194]

[0195] In the specific calculation of f in (i,j,k) and f out (i,j,k), it is necessary to determine according to the decision variables x and corresponding flow weights of the adjacent grid entering and leaving the current grid, and examples are as follows:

[0196]

[0197] where and represent the flow weights from different adjacent grids to the current grid and from the current grid to different adjacent grids, respectively. These weights can be set according to factors such as road width, traffic capacity, and characteristics of different areas in the actual chemical park.

[0198] (3) Objective function

[0199] 1) Shortest path length objective

[0200] If the objective is to minimize the total length of the emergency rescue path and the emergency evacuation path, the following objective function can be constructed:

[0201]

[0202] The meaning is that for all possible adjacent grid pairs, the distance between them is calculated and multiplied by the corresponding decision variable (only when both grids are on the path, the product of this term will contribute to the total length), and finally all these products are summed up, and the result is the total length of the entire emergency rescue and emergency evacuation path, the goal is to minimize this total length.

[0203] 2) Shortest time objective

[0204] When the total time of emergency rescue and emergency evacuation is expected to be minimized, the objective function can be set as:

[0205]

[0206] Here The calculation of (travel time) is usually determined according to the distance and the average travel speed of the corresponding road segment, and the specific calculation formula can be expressed as:

[0207]

[0208] Where represents the average travel speed (in units such as meters / second, etc.) from grid G i,j,k to grid The travel speed of different areas (such as flat road areas, areas with obstacles, areas covered with dangerous substances, etc.) is different, and needs to be set according to the specific situation of each area in the actual chemical industrial park.

[0209] 3) Determine the travel cost by considering multiple factors such as distance, danger level (for example, measure the danger level of different areas according to factors such as dangerous substance concentration, high temperature area of fire, etc.), congestion, etc. When the objective function is to minimize the total travel cost:

[0210]

[0211] The determination of travel cost can be achieved by constructing a cost function, for example:

[0212]

[0213] Where represents the average travel speed (in units such as meters / second, etc.) from grid G to grid The level of danger (which can be quantified based on actual relevant data such as the concentration of hazardous substances and temperature) The index representing the congestion level of the road segment (which can be obtained through real-time monitoring data or estimation) includes α, β, and γ, which are corresponding weighting coefficients used to adjust the proportion of different factors in the total cost. Their values ​​can be reasonably set according to the environmental characteristics of the actual chemical industrial park and the focus of emergency decision-making.

[0214] (4) Ant Colony Algorithm

[0215] The mathematical model for two-way path planning in emergency rescue and evacuation under a three-dimensional environment, as described above, typically falls under the category of integer programming problems (because the decision variable x...). i,j,k The value can only be 0 or 1.

[0216] For smaller-scale problems, precise algorithms, such as branch and bound or cutting plane methods, can be used to find the optimal path planning solution. However, in real-world chemical industrial park scenarios, the number of grid cells is often extremely large, making precise algorithms computationally inefficient and unable to meet the real-time requirements of emergency response phases.

[0217] Therefore, the heuristic ant colony algorithm is adopted in practical applications. Through continuous iterative search, a better path planning scheme is found within an acceptable time range, thereby providing effective path guidance for emergency rescue and evacuation work in the emergency response phase of chemical industrial parks, and ensuring that rescue and evacuation operations can be carried out efficiently and safely.

[0218] This comprehensive and detailed bidirectional path planning mathematical model fully considers various complex factors in emergency scenarios in chemical industrial parks under three-dimensional conditions. Combined with the corresponding objective function, constraints, and appropriate solution methods, it helps to plan scientific, reasonable, and feasible emergency rescue and evacuation paths.

[0219] 5.3 Dynamic Environment Intelligent Obstacle Avoidance Model

[0220] (1) Model basis and definition of relevant variables

[0221] 1) Spatial and Raster Representation

[0222] The three-dimensional space of the chemical industrial park is divided into a series of cubic grids G. ijk (i, j, and k represent row, column, and level indices, respectively). Each grid cell has multiple attributes, such as whether it is an obstacle (using o). ijk It means, o ijk =1 indicates an obstacle, o ijk =0 indicates that it is not an obstacle), and the concentration of hazardous substances c ijk Temperature t ijkEtc. These attributes will affect the navigation and obstacle avoidance decisions of agents (e.g. rescue vehicles, evacuees, etc.).

[0223] 2) Agent-related variables

[0224] Let the current position of the agent be (i α ,j α ,k α ), and the target position be (i t ,j t ,k t ). The velocity vector of the agent is represented as where v x , v y , v z are the velocity components of the agent in x, y, z directions respectively.

[0225] 3) Sensor perception range and angle

[0226] Suppose the agent is equipped with a sensor, whose perception range is R (unit: meter or other length unit) in radius, and the perception angle range is [a min , a max ] in horizontal direction, and [b min , b max ] in vertical direction. This means that the agent can only obtain the environmental information within a certain angle and distance range to make obstacle avoidance decisions.

[0227] (2) Dynamic obstacle avoidance adjustment combined with path planning

[0228] 1) Path re-planning trigger condition

[0229] During the agent's travel along the planned path, when new obstacles are encountered, the dangerous area changes, or other conditions that make the original path impassable, path re-planning needs to be triggered. For example, a threshold can be set, when the distance between the agent and the newly appeared obstacle is less than the threshold and the original path cannot be avoided in the direction, or the concentration of dangerous substances exceeds the safety standard making the original path unsafe, the path re-planning mechanism is started.

[0230] 2) Path re-planning and obstacle avoidance coordination

[0231] Once the re-planning is triggered, the current obstacle avoidance situation and environmental information need to be combined to re-plan the path. The bidirectional path planning mathematical modeling method mentioned earlier can be used, but at this time the position, speed and surrounding obstacle avoidance related information of the current agent should be considered as constraint conditions. For example, the obstacle avoidance related cost term is added in the objective function of path planning, so that the newly planned path not only meets the distance, time and other conventional optimization goals, but also tries to avoid high-risk obstacle areas and dangerous areas. The objective function can be modified as follows (taking the minimum comprehensive cost as an example, adding the obstacle avoidance cost on the basis of the original):

[0232]

[0233] Wherein c avoid (i,j,k,i2j2k2) represents the obstacle avoidance cost from the grid G ijk to the grid , which can be determined according to factors such as the obstacle density and dangerous substance concentration of the area. Through such a cooperative mechanism, the agent (rescue vehicle, evacuated personnel, etc.) can continuously and effectively avoid obstacles and dangerous areas in the dynamic environment and move towards the target.

[0234] Through the intelligent obstacle avoidance model composed of the above formulas and mechanisms, the agent (rescue vehicle, evacuated personnel, etc.) can effectively avoid obstacles, dangerous areas, etc. in the complex three-dimensional dynamic environment of the chemical industry park, and ensure the smooth progress of emergency rescue and emergency evacuation.

[0235] 6. Scientific evaluation system

[0236] Real-time collection and analysis of exercise data, including response time of emergency rescue forces, rationality of resource allocation, effect of accident control, etc. Scientific evaluation standards and index system are developed. By comparing with the preset evaluation standards and index system, the evaluation of each link and the overall effect of the exercise is carried out. Feedback information is provided to the exercise organizers and participants in a timely manner, and the problems and deficiencies in the exercise are pointed out, and suggestions for improvement are put forward. For example, if it is found that the response time of the emergency rescue team is too long, it may be due to unreasonable traffic route planning or poor dispatching command. The feedback module will prompt the exercise organizers to adjust the rescue route or optimize the dispatching strategy. At the same time, the exercise organizers can adjust the exercise scene or emergency response strategy in real time during the exercise according to the feedback information, such as increasing rescue forces, changing rescue tactics, etc., to observe the influence of different strategies on the exercise results, and realize the dynamic optimization of the exercise.

[0237] In summary, the present invention creates a high-precision digital twin model, Internet of Things device data fusion processing, accurate simulation of emergency scenarios, efficient scheduling of emergency resources, rescue and evacuation path planning, real-time response and evaluation, etc. Technical solutions comprehensively improve the emergency response capability and safety management level of the chemical industry park.

[0238] The digital twin-based chemical industrial park emergency processing method of the present application can construct multi-level, multi-dimensional and multi-fine-grained chemical industrial park emergency event disposal models in different scenarios, making more alternative solutions or decision theories that cannot be verified due to limited physical space and reliance on real physical entities possible through continuous trial and error and optimization.

[0239] (1) Significantly improve the realism and accuracy of emergency simulation

[0240] Accurately simulate the development process of various emergency events in the chemical industrial park, including fire spread, toxic gas diffusion, personnel evacuation, etc., to improve the realism and accuracy of simulation drills.

[0241] (2) Significantly reduce the cost of emergency simulation drills

[0242] The present application can realize comprehensive simulation and drilling of emergency events without investing a large amount of site, equipment and manpower costs. By repeatedly using simulation scenarios and data, the drilling cost is significantly reduced, while the efficiency and effectiveness of the drilling are improved.

[0243] (3) Significantly enhance the real-time and effectiveness of emergency response

[0244] With the aid of visualization technology and data analysis tools, through the data acquisition and transmission system, the present application can monitor and respond to emergency events occurring in the park in real time, intuitively display the development process and impact range of emergency events, and provide intuitive and clear information support for emergency decision-making.

[0245] The chemical industrial park comprehensive emergency simulation and drilling system and method based on high-precision digital twin technology takes high-precision digital twin technology as the core, combines advanced technologies such as Internet of Things, big data, cloud computing, etc., and builds a virtual simulation environment corresponding to the physical environment of the chemical industrial park, which can simulate various emergency scenarios and provide strong support for emergency management and safety training in the chemical industrial park.

[0246] Build a high-precision digital twin model: High-precision digital modeling of the physical environment of the chemical industrial park, accurate restoration of equipment, facilities, buildings, etc. in the park, and real-time perception and reflection of dynamic changes in the park, such as equipment operating status, personnel flow, etc.

[0247] Scenario simulation: Provide a variety of chemical industrial park emergency scenarios, including fire, explosion, chemical leakage, etc., which can be selected according to actual needs to conduct drills.

[0248] Resource scheduling: Construct an emergency resource scheduling model for chemical industry park accidents, integrate digital twin simulation technology and hybrid intelligent optimization algorithm, realize efficient allocation of resources (such as fire fighting equipment, medical rescue materials, protective equipment, etc.) in dynamic and complex environment, significantly improve the response speed, resource utilization and efficiency of accident response.

[0249] Rescue path planning: Considering the necessary points and bypassing areas, the target object can find a collision-free and safe path from the starting point to the ending point within the specified range of the park.

[0250] Evaluation and feedback: After the drill, evaluate the drill process and generate an evaluation report, including drill time, task completion, personnel performance, etc. Users can summarize lessons learned and improve emergency plans based on the evaluation report.

Claims

1. A chemical industrial park emergency treatment method based on digital twinning, characterized in that Comprise the following steps: Step S1: Constructing a high-precision digital twin model: high-precision digital modeling of the entity environment of the chemical industry park is carried out to realize accurate restoration of equipment, facilities and buildings in the park, and to realize real-time sensing and reflection of dynamic changes in the park; Step S2: Fusion of Internet of Things device model and twin scene, matching the Internet of Things device model with the actual park enterprise device information to realize the cooperative linkage of the model and the monitoring device; Step S3: Provide multiple chemical industry park emergency scene simulation, at least including fire, explosion, chemical leakage, and select different scenes for drilling according to actual needs; Step S4: Constructing a chemical industry park accident emergency resource scheduling model, fusing digital twin simulation technology and hybrid intelligent optimization algorithm to realize resource allocation in dynamic and complex environment; Step S5: Considering the necessary points and bypass areas, so that the target object finds a collision-free safe path from the starting point to the ending point within the park specified range; The emergency resource scheduling model of step S4 comprises constructing an interaction mechanism between the agent and the digital twin scene, and the specific process is as follows: Step S41-1: Interaction mechanism between resource supply agent and digital twin scene: Obtain accurate resource reserve information of resource supply point in real time through digital twin scene, at least including the number, storage location and validity period of each resource; at the same time, according to the prediction of resource consumption trend of digital twin model, combine with reinforcement learning algorithm to decide resource allocation plan; Step S41-2: Interaction mechanism between transportation agent and digital twin scene: Based on the dynamic information of road network in digital twin scene, plan the optimal transportation route and adjust the transportation strategy; the interaction between transportation agent and digital twin scene also includes real-time simulation of transportation process, prediction of possible problems in different transportation schemes and avoidance in advance; Step S41-3: Interaction mechanism between rescue site agent and digital twin scene: Make full use of digital twin scene to perceive the detailed situation of the accident site, through deep interaction with digital twin scene, the rescue site agent can more accurately evaluate the resource demand, and provide more targeted resource demand adjustment suggestions to other agents.

2. The digital-twin-based chemical industrial park emergency treatment method according to claim 1, characterized in that ； Step S41-1 comprises: Resource reserve information in digital twin scenarios: including the current actual quantity of various resources at various supply points Storage location coordinates; State space: Wherein K is the total number of resource types; Action space: Resource types provisioned by each provisioning point and quantities Resource replenishment time information i.e. the estimated time needed from the moment the resource replenishment request is issued until the resource is actually replenished in place; Resource consumption rate predicted by digital twin The rate of consumption of each resource per unit of time is predicted based on factors of the type, size, and development trend of the incident; Reward function R: 1) Rescue time reward function: wherein T windows is the optimal rescue time window for the accident simulated by the digital twin, is the completion time of the i-th rescue sub-task; 2) Resource utilization rate reward function: wherein is the kth resource requirement for digital twin prediction, and δ is the allowable error range. Using reinforcement learning algorithm, the agent selects action A θ (a|s) according to the current state S supply through the policy network π supply , and updates learning according to the reward function R, the goal is to maximize the long-term cumulative reward γ is the discount factor; Step S41-2 comprises: Road network information G=(V,E) in digital twin scene: including nodes V and edges E, and the basic attributes of each edge; Real-time traffic condition C ij : Provided by the fusion of the Internet of Things sensing device and the digital twin scene, including the congestion degree, the passing speed v ij of each road, the congestion degree is represented by the congestion coefficient ρ ij , 0≤ρ ij≤1 , 0 represents smooth, 1 represents complete congestion; Position L of the transport means t and status S t : real-time geographical position coordinates of the transport vehicle acquired by the Internet of Things perception device and the operating status of the vehicle; Real-time weather conditions W in a digital twin scenario condition : including wind speed ω speed , wind direction ω direction , rainfall intensity r intensity , which can affect transport safety and efficiency; Potential interference of other associated incidents in the perimeter area on the transport route A interferense ; State space: S transport = {G, C ij , L t , S t , W condition , A interferense}; Action space: A transport = {Route ij , Veh m icle adjust} ; Reward function: Transportation cost reward function where C transport =∑ i (C fuel-i +C depreciation-i +C labor-i ), C fuel-i = S ij × c fuel × (1+α ω ×ω speed +α r ×r intensity ); where c fuel is the fuel consumption cost per unit distance, α ω and α r are coefficients of the impact of wind speed and rainfall intensity on fuel consumption, ΔC transport is the cost increase due to dynamic changes in road conditions, and C budget is the preset transportation cost budget. An enhanced learning algorithm is adopted to learn the optimal transportation strategy according to S transport State selection action A transport And learning update is performed according to the reward function to optimize the transportation strategy. Step S41-3 comprises: Incident information in the digital twin scenario: including the incident type T accident , real-time development information of the incident; On-site resource usage: types of resources that have arrived at the incident site and the number of each progress of resource usage; Digital twin scenario simulation accident development trend prediction T accident-trend ; Resource demand adjustment proposals: Proposals to the resource supply agent and the transportation agent of the kind of resources that need to be allocated and the amount of the adjustment proposal; State space: wherein the severity of the accident S severity According to the index quantification evaluation of the accident influence range and the dangerous substance leakage amount, the rescue progress P rescue By the proportion of the completed rescue tasks or the key indicators Action space: Reward function: Accident loss reward function where L accident-predict is the real-time assessment and prediction of the accident loss by the digital twin scenario, including the prediction of the personnel casualty L casaultu-predict , the prediction of the property loss L property-predict and the prediction of the environmental damage L environment-predict , L max is the acceptable maximum accident loss threshold; Based on reinforcement learning algorithm, according to state S scene Generate action A scene And learn optimization according to reward function to improve rescue efficiency and reduce accident loss.

3. The digital-twin-based chemical industrial park emergency treatment method according to claim 2, characterized in that In the above step S4, the digital twin simulation technology and the hybrid intelligent optimization algorithm are fused, which comprises establishing a digital twin enabled network flow optimization module; the specific process is as follows: Step S42-1: Network construction and dynamic parameter update Resource supply point information in the digital twin scene: including the position coordinates of the supply point, the upper limit of the resource reserve And the supply capacity decline ratio β caused by the influence of surrounding accidents i , β i >0; Accident site information: accident location coordinates, dynamically fluctuating resource demand predictions of the digital twin scenario Road network information: including a set of network nodes V' and a set of edges E', and the dynamic changes of road capacity monitored by the digital twin scenario and real-time adjustment of transport time Updated network flow model: including adjusted set of network nodes V', set of edges E', edge capacity matrix C', transit time matrix T' and resource demand vector Q required ; Model formula and logical relationship: A network N = (V', E', C', T', D') is constructed; wherein D' is a digital twin scenario driven dynamic parameter matrix containing and information; Step S42-2: Dynamic adjustment of optimization target and constraint condition a degree of reduction R of the accident risk assessed by the digital twin scenario risk-reduce : a reduction ratio a of the accident spread range simulated by the digital twin scenario spread-reduce , a reduction ratio a of the dangerous substance concentration concentration-reduce quantitative assessment of the index Final output optimized resource scheduling scheme: determine the resource transportation flow on each transportation route Final allocation of resource supply points to meet optimization objectives and constraints; Model formula and logical relationship: Objective function Z = λC total + (1 - λ)T total + μR risk-reduce ; where, Compute T total = ∑ (i,j)∈∈ T' ij x ij ; Constraint condition: 1) Resource provisioning constraints: 2) Resource requirement constraints: 3) Flow conservation constraint: for transit nodes k∈V' except resource supply points and incident sites, ∑ i:(i,k)∈E' x ik ≤∑ j:(k,j)∈E' x kj In consideration of the new nodes or node failure that may occur in the digital twin scene, the calculation range and node relationship are dynamically adjusted. 4) Capacity limit constraints: Adopting adaptive network flow optimization algorithm, the optimal solution is gradually approached by constantly finding augmenting path and adjusting network flow to meet the objective function and constraint condition.

4. The digital-twin-based chemical industrial park emergency treatment method according to claim 3, characterized in that The fusion of digital twin simulation technology and hybrid intelligent optimization algorithm in step S4 further includes establishing a deep coordination mechanism between the digital twin and the agent and the network flow; the specific process is as follows: Step S43-1: Real-time data sharing and synchronization Real-time data in the digital twin scene: including every second updated status data of the accident site, real-time traffic condition data of the road network, and every second updated resource reserve and consumption data of the resource supply point; Decision information of multi-agent: resource allocation decision of resource supply agent and Transport route selection Route of transport agent ij , Vehicle selection Veh icle m and transport speed adjustment strategy V adjust Resource demand adjustment suggestion of rescue site agent and Optimization result of the network flow optimization module: optimized resource scheduling scheme, i.e., resource transportation flow on each transportation route Push the data in the digital twin scene to the multi-agent reinforcement learning module and the network flow optimization module respectively; Deliver the multi-agent decision information to the digital twin scene for simulation evaluation, and synchronize the evaluation results to the network flow optimization module; Feedback the optimization results of the network flow optimization module to the digital twin scene, and then deliver them to the multi-agent; Step S43-2: Collaborative decision-making and dynamic adjustment Synchronization data and evaluation results of the digital twin scene: simulation evaluation of the multi-agent decision effect, and simulation verification results of the network flow optimization scheme in the digital twin scene; Current state information of multi-agent: including resource supply agent's resource reserve change state S supply , transport process state S of transport agent transport , the latest state S of the accident site of the rescue site agent scene ; Optimization scheme information of the network flow optimization module: resource transportation flow allocation Resource supply point deployment plan; Output dynamic adjustment decision: multi-agent adjusts its own decision according to the feedback of digital twin scene and network flow optimization result, including resource supply agent adjusting resource allocation and type and Transport agent adjusts transport route Route ij-new , transport speed V adjust-new , rescue site agent adjusts resource demand suggestion and Network flow optimization module further optimizes resource scheduling scheme according to the new decision of multi-agent and the dynamic change of digital twin scene, and generates new resource transportation flow In the multi-agent reinforcement learning module, each agent recalculates the action selection strategy according to the new input state and the feedback information obtained from the digital twin scene and the network flow optimization module; the strategy update formula of the resource supply agent is expressed as: where θ supply is a policy network parameter of the resource supply agent, α supply is a learning rate, is a policy function, is a new resource allocation action, is a new state, is a new reward; The network flow optimization module re-builds the network flow model and performs optimization calculation according to the new decisions of the multi-agent and the dynamic parameter changes in the digital twin scene; the network flow optimization module adjusts the edge capacity and transportation time parameters related to the route accordingly, and re-solves the objective function: wherein total cost, total time and accident risk reduction degree calculated based on the new network flow model and the accident development, respectively, and the new resource transportation flow is obtained by solving to achieve dynamic optimization of the resource scheduling scheme.

5. The digital-twin-based chemical industrial park emergency treatment method according to claim 1, characterized in that The above step S5 finds a collision-free safe path, including emergency rescue and evacuation path planning, and the specific process is as follows: Rescue / evacuation path planning: considering the must-pass points and bypass areas, a collision-free safe path from the starting point to the ending point is found for the target object within the specified range of the park; Step S51-1: Dynamic grid method environment modeling The dynamic grid method is used for environment modeling, which divides the environment of the chemical industry park into small square cells, i.e. grids; each grid represents a region in the environment and provides information about obstacles and free space; the environment map is gradually constructed by updating the state of the grid; Step S51-2: Representation and initialization of grid Each grid has its corresponding state attribute, which is defined as three basic states: idle, obstacle, and unknown; Let S(G ijk ) denote the state of the grid G ijk , which is initialized by At the beginning of the chemical industrial park environment modeling, for those obviously known road area corresponding grid, S(G ijk ) = 0; for the building, large chemical equipment occupied area corresponding grid, S(G ijk ) = 1; and those have not been detected by sensor corresponding grid, S(G ijk ) = -1; The three-dimensional grid map M of the entire chemical industry park is represented as the set of all grid states, i.e. M = {S(G ijk )|i = 1,2,...,I; j = 1,2,...,J; k = 1,2,...,K} Where I, J, and K are the total number of grid maps in the row, column, and layer dimensions respectively; Step S51-3: Grid state update based on sensor information Assuming that the sensor detects the environment, the detection results have certain uncertainty, and the Bayesian filter is used to update the probability estimate of the grid state; Let P(S(G ijk ) = s | Z 1:t ) denote the probability that the grid G 1:t is in state s after receiving a sequence of sensor observations Z ijk from time 1 to time t; s takes values 0, 1, or -1. According to the Bayesian formula, the update process is as follows: 1) Prediction step, i.e. prior probability update: Here P(S(G ijk ) = s | S(G ijk ) = s', Z 1:t-1 ) denotes the transition probability of the grid from state s' at the previous time step to state s at the current time step given the previous observation, and depends on the dynamic properties of the environment and the performance factors of the sensor. P(S(G ijk ) = s'| Z 1:t-1 ) is the probability of the previous grid being in state s' from the update result of the last round; 2) Update step, i.e. posterior probability update combined with current observation: Where P(Z) t |S(G ijk ) = s) indicates that in the raster G ijk When in state s, obtain the observation value Z at the current time t. t The probability is determined by the sensor's observation model; Step S51-4: Grid map construction and visualization In order to visualize the grid map in three-dimensional space, it is necessary to convert the row and column layer coordinates of the grid into actual three-dimensional space coordinates; let the edge length of the grid in the x, y, z direction be d x , d y , d z , the lower left corner coordinates of the grid G ijk (x ijk , y ijk , z ijk ) are: x ijk = j x d x y ijk = i x d y z ijk = (k - 1) x d z Set the grid G ijk with the center coordinates (x ijk ,y ijk ,z ijk ) of the formula as follows:

6. The digital-twin-based chemical industrial park emergency treatment method according to claim 5, characterized in that The safe path without collision found in the step S5 further includes two-way path planning mathematical modeling, and the emergency response is divided into two stages of internal emergency response and inter- and intra-garden collaborative emergency response; the target is to plan the optimal two-way path for emergency rescue and emergency evacuation in different stages, so as to realize the efficiency of rescue and evacuation, and avoid road conflict and congestion; the specific process is as follows: Step S52-1: model basic elements and variable definition 1) space representation and grid definition Assume that the three-dimensional space where the chemical industrial park is located is discretized into a series of cubic-shaped grids, denoted as G ijk where i, j, k are all positive integers. 2) decision variable setting Introducing decision variables x ijk , to represent the path planning case; If a path passes through a grid G ijk then x ijk = 1; If the path does not pass through the grid, then x ijk = 0; 3) start point and end point definition Emergency rescue: set the emergency rescue starting point coordinates as (i r ,j r ,k r ), and the corresponding rescue target end coordinates as Emergency evacuation: the starting point coordinate of emergency evacuation is set as (i e ,j e ,k e ), and the ending point coordinate of evacuation target is 4) definition of other related parameters Distance metric: Let d(x, y) denote the distance from grid point x to grid point y. ;​​ Travel time: denotes the time required to travel from grid to grid ; Passing cost: The passing cost is used to measure the comprehensive passing cost from the grid to the grid , and the specific value is determined by considering the factors of distance, concentration of dangerous substances, and road congestion degree. Step S52-2: constraint condition 1) path connectivity constraint In three-dimensional space, connectivity needs to be maintained between the grids on the path, i.e. for any two grids that are successively passed through and They should be spatially adjacent; for a three-dimensional grid, a grid has 26 neighboring grids, i.e. including the grid itself, the grids on the same plane, and the grids on the upper and lower planes. A neighborhood function N(G i,j,k ) is defined to represent all the neighborhood grid sets of grid G i,j,k , and the mathematical formula of the path connectivity constraint is represented as: The meaning of this formula is that if a certain grid... On the planned path, that is Then at least one of its neighboring rasters must also be on the path, that is, the corresponding neighboring raster. The sum of these values ​​must be at least 1 to ensure that the planned path is continuous and passable; 2) start point and end point constraint For the emergency rescue path and the emergency evacuation path, there are definite start point and end point requirements, and the corresponding constraint formula is as follows: Emergency rescue path start point constraint: The rescue path must start from the set rescue start point, so the decision variable x corresponding to the start point grid is assigned a value of 1; Emergency rescue path end point constraint: Ensure that the rescue path can finally reach the set rescue target end point, so the decision variable x corresponding to the end point grid is also assigned a value of 1; For the emergency evacuation path, the emergency evacuation path start point constraint is: The emergency evacuation path end point constraint is: 3) flow conservation constraint, that is, considering the case of multiple subjects traveling at the same time When there are multiple rescue teams or multiple batches of evacuation personnel planning paths at the same time, in order to ensure flow balance, the flow entering a grid is equal to the flow leaving the grid; Let f in (i,j,k) represents the flow rate into the grid, and (i,j,k) represents the flow rate out of the grid. ijk If the flow rate is such that the flow conservation constraint is expressed as follows: In the specific calculation of f in (i,j,k) and f out (i,j,k), it is necessary to determine according to the decision variables x passing through the neighborhood grid into and out of the current grid and the corresponding flow weight factors: wherein and respectively represent the flow weight from different neighborhood grids to the current grid and from the current grid to different neighborhood grids; Step S52-3: objective function 1) shortest path length objective If the total length of the emergency rescue path and the emergency evacuation path is minimized as the objective, the following objective function is constructed: The meaning is that for all possible adjacent grid pairs, the distance between them is calculated and multiplied by the corresponding decision variable, and finally all these products are summed up, and the result is the total length of the entire emergency rescue and emergency evacuation path, and the target is to minimize the total length; 2) shortest time objective When the total time of emergency rescue and emergency evacuation is expected to be minimized, the objective function is set as: Travel time The calculation of the travel time is determined according to the distance and the average travel speed of the corresponding road section, and the specific calculation formula is expressed as: wherein represents the average travel speed from the grid G i,j,k to the grid G; 3) Determine the passing cost by considering the distance, the dangerous degree, and the congestion When the total passing cost is minimized as the objective function: Cost of passage The determination is achieved by constructing a cost function: wherein represents a risk level index from the grid to the grid , and represents a congestion level index of the road section, and α, β, γ are corresponding weight coefficients for adjusting the proportion of different factors in the total cost.

7. The digital-twin-based chemical industrial park emergency treatment method according to claim 6, characterized in that The safe path without collision found in the step S5 further includes establishing a dynamic environment intelligent obstacle avoidance model, and the specific process is as follows: Step S53-1: model basis and related variable definition 1) space and grid representation The three-dimensional space of the chemical industrial park is divided into a series of cuboid-shaped grids G ijk , i, j, k represent row, column, and layer indexes, respectively, and each grid has multiple attributes; 2) agent related variables Let the current position of the agent be (i α ,j α ,k α ), the target position be (i t ,j t ,k t ), and the velocity vector of the agent be where v x , v y , v z are the velocity components of the agent in the x, y, z directions, respectively. 3) sensor sensing range and angle Assume that the agent is equipped with a sensor whose perception range is R in radius, whose perception angle range is [α min ,α max ] in horizontal direction, and [β min ,β max ] in vertical direction; Step S53-2: dynamic obstacle avoidance adjustment combined with path planning 1) path re-planning trigger condition When the agent travels along the planned path, if a new obstacle, a change in dangerous area or other conditions that make the original path impassable is encountered, path re-planning needs to be triggered; a threshold is set, when the distance between the agent and the newly appeared obstacle is less than the threshold and the original path cannot be avoided in the direction, or the concentration of dangerous substances exceeds the safety standard and makes the original path unsafe, the path re-planning mechanism is started; 2) path re-planning and obstacle avoidance coordination Once the re-planning is triggered, the path planning needs to be re-done with the current obstacle avoidance situation and environmental information. The bidirectional path planning mathematical modeling method is adopted, but the position, speed of the current agent and the surrounding obstacle avoidance information are considered as constraint conditions at this time. The objective function is modified as follows. Taking the minimum cost minimization goal as an example, the obstacle avoidance cost is added to the original basis: where c avoid (i,j,k,i2j2k2) represents the obstacle avoidance cost from the grid G ijk to the grid , which is determined according to the factors of the obstacle density, the dangerous substance concentration of the region, and through such a synergistic mechanism, the agent can continuously and effectively avoid obstacles and move towards the target in a dynamic environment.

Images