Tunnel fire-fighting emergency linkage method and system based on digital twinning

Abstract

The invention relates to the technical field of tunnel safety and emergency management, and discloses a tunnel fire-fighting emergency linkage method and system based on digital twinning, and the method comprises the steps: collecting the multi-modal data of a fire environment through a plurality of sensors, and constructing a global situation understanding model; processing the environment state and the crowd behavior characteristics by adopting a multi-head attention network; dynamically adjusting the attention weight of the crowd feature dimension by applying a situation-enhanced attention mechanism; constructing a psychological environment interaction model, and predicting crowd behavior evolution in different fire scenes; generating a situation-adaptive multi-level evacuation guidance strategy tree; real-time effect evaluation is carried out, and system overall optimization is realized through a digital twinborn model; according to the method, the environment state and the crowd behavior characteristics are processed at the same time through the multi-head attention network, the internal correlation between the environment factors and the crowd behaviors can be understood, and therefore comprehensive and accurate cognition is formed on the tunnel fire scene.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel safety and emergency management, and more specifically, to a tunnel fire emergency linkage method and system based on digital twins. Background Art

[0002] With the acceleration of urbanization, tunnels are being widely constructed and used as important transportation infrastructure. However, once a tunnel fire accident occurs, due to its enclosed nature, it often causes serious casualties and property losses. Currently, tunnel fire emergency systems still face many technical challenges in practical application:

[0003] Existing technical solutions generally suffer from a single perception capability. Traditional systems typically focus solely on monitoring physical fire parameters (such as temperature and smoke concentration) and are unable to simultaneously integrate information on the fire situation and the psychological state of the crowd. This results in an incomplete understanding of complex fire scenarios and an inability to fully grasp the complex relationship between fire development and crowd response. Existing evacuation guidance strategies mostly use preset plans that are disconnected from actual situations. These systems struggle to provide differentiated guidance based on the dynamic changes in fire conditions and crowd psychological states in different areas, and are unable to adapt to the uncertainty of fire development and the complexity of crowd behavior. Existing technologies lack an understanding of the evolution of crowd behavior in fire scenarios. During emergency evacuations, crowd behavior is influenced by multiple factors, including environmental threats, individual psychological states, and group effects. Existing systems have limited predictive capabilities for these complex behavioral patterns, which can lead to misjudgments at critical moments and affect evacuation effectiveness. Existing systems generally lack collaborative perception and analysis mechanisms that can simultaneously consider environmental factors and crowd psychological states. The complex interactive relationship between the fire environment and crowd psychological states exists, and traditional systems are unable to effectively capture and utilize this interactive relationship to optimize emergency decision-making.

[0004] The above technical problems make it difficult for existing systems to provide accurate and effective emergency guidance in complex tunnel fire evacuation scenarios, increasing the risk of casualties. There is an urgent need for a new tunnel fire emergency linkage method that can comprehensively consider the environmental situation and the psychological state of the crowd. Summary of the Invention

[0005] The present invention provides a tunnel fire emergency linkage method and system based on digital twins, which solves the technical problems in related technologies that are unable to collaboratively perceive the fire situation and the psychological state of the crowd and difficult to provide differentiated evacuation guidance.

[0006] The present invention provides a tunnel fire emergency linkage method based on digital twins, comprising:

[0007] Collect multimodal data of the fire environment through various sensors and build a global situation understanding model;

[0008] On the basis of the global situation understanding model, a multi-head attention network is used to process environmental state and crowd behavior characteristics to realize the collaborative perception of environment and crowd information.

[0009] Combined with the characteristics of fire scenes, a context-enhanced attention mechanism is applied to dynamically adjust the attention weight of crowd feature dimensions.

[0010] Based on the adjusted attention weight, a psychological environment interaction model is constructed to predict the evolution of crowd behavior under different fire scenarios.

[0011] According to the prediction results, a multi-level evacuation guidance strategy tree is generated to provide differentiated evacuation guidance.

[0012] During the implementation of the evacuation guidance strategy, real-time effect evaluation is carried out, and the overall system optimization is realized through the digital twin model.

[0013] Further, the global situation understanding model includes:

[0014] A multi-modal encoder unit is used to process different modal sensor data.

[0015] A cross-modal attention unit is used to calculate the correlation weight matrix between different modalities.

[0016] A feature fusion unit is used to weight and fuse each modal feature according to the attention weight.

[0017] A situation representation unit is used to convert the fused features into a physical quantity distribution representation of the tunnel space.

[0018] Further, the multi-head attention network includes:

[0019] An environmental feature input unit is used to receive and process environmental state features in the global situation understanding model.

[0020] A crowd feature input unit is used to receive and process crowd behavior characteristics.

[0021] A multi-head attention calculation unit includes Attention heads, each responsible for focusing on different environmental and crowd interaction patterns.

[0022] An attention output fusion unit is used to weight and fuse the outputs of Attention heads.

[0023] Further, the psychological environment interaction model includes:

[0024] An environmental perception unit is used to receive environmental state parameters.

[0025] Psychological state modeling unit, used to model the psychological state of the crowd;

[0026] Behavior evolution prediction unit, used to predict changes in crowd behavior;

[0027] The dangerous state identification unit is used to identify the dangerous state.

[0028] Furthermore, the multi-level evacuation guidance strategy tree includes:

[0029] The root node unit represents the starting point of the overall evacuation decision;

[0030] Environmental state branch unit, used to divide the tunnel space into different area nodes;

[0031] Psychological state branching unit, used for further branching based on the psychological state of the crowd;

[0032] The guidance strategy leaf node unit contains the specific guidance strategy for a specific combination of environmental state and psychological state;

[0033] The priority scoring unit is used to assign a priority to each guidance strategy.

[0034] Furthermore, the context-enhanced attention mechanism includes:

[0035] Obtain fire scene characteristic data, including fire location, scale, spread speed, and smoke diffusion status;

[0036] Extract and encode the fire scene characteristics and convert them into scene characteristic vector representation;

[0037] According to the scene feature vector, calculate the attention weights for different crowd feature dimensions;

[0038] Apply the calculated attention weights to the multi-head attention network to dynamically adjust the attention paid to different demographic characteristics.

[0039] Based on the adjusted attention allocation, the importance of crowd behavior characteristics is scored to highlight the most critical features for evacuation guidance in the current scenario.

[0040] Furthermore, the real-time effect evaluation includes:

[0041] Define evaluation indicators for guidance effectiveness, including evacuation speed, smoothness of crowd flow, and panic control;

[0042] Collect real-time data through sensor networks to monitor the effectiveness of guiding strategy execution;

[0043] The deviation function is used to calculate the deviation between the measured effect and the expected effect;

[0044] Generate parameter adjustments based on the deviation analysis results.

[0045] Furthermore, the digital twin model includes:

[0046] A data synchronization unit, used to receive and process the situation data acquired in real time;

[0047] A virtual mapping unit, configured to map physical entity information to a virtual space;

[0048] 3D visualization unit, used to display fire situation and crowd distribution;

[0049] Historical data storage unit, used to save historical data of system operation;

[0050] Parameter optimization unit, used to optimize the parameters of each module of the system.

[0051] Furthermore, the generation of the scenario-adaptive multi-level evacuation guidance strategy tree uses a multi-objective optimization algorithm, which includes:

[0052] Define three optimization goals: shortest evacuation time, minimum congestion risk, and lowest psychological stress;

[0053] Construct a set of constraints, including the number of evacuation channels and the number of guidance equipment;

[0054] Use Pareto optimization method to solve multi-objective optimization problems;

[0055] According to the characteristics of the scenario, the most suitable evacuation plan is selected from the optimal solution set.

[0056] The present invention provides a tunnel fire emergency linkage system based on digital twins, which is used to implement the above-mentioned tunnel fire emergency linkage method based on digital twins, including:

[0057] Situational awareness module, used to obtain fire environment data through sensor networks and build an overall situation;

[0058] Collaborative analysis module, used to process environmental and crowd characteristic information and achieve multi-dimensional information fusion;

[0059] Adaptive attention module, used to dynamically assign attention priorities based on fire development characteristics;

[0060] Behavior prediction module, used to build psychological response and behavior models of people under different environmental conditions;

[0061] Strategy generation module, used to create differentiated evacuation guidance plans based on scene characteristics and crowd status;

[0062] The system optimization module is used to evaluate the guidance effect and realize dynamic parameter adjustment through virtual-real mapping.

[0063] The beneficial effects of the present invention are: by simultaneously processing environmental conditions and crowd behavior characteristics through a multi-head attention network, the system can understand the inherent relationship between environmental factors and crowd behavior, thereby forming a comprehensive and accurate understanding of the tunnel fire scene;

[0064] Through the context-enhanced attention mechanism, the system can dynamically adjust the attention weights on different characteristic dimensions of the crowd according to the characteristics of the fire, enabling the system to adaptively focus on the most critical crowd behavior characteristics in the current scene, avoiding the blind spot problem caused by fixed attention allocation in traditional systems;

[0065] The mind-environment interaction model enables the system to predict the evolution of crowd behavior in different fire scenarios and identify potential dangerous conditions in advance. This solves the problem of traditional systems' insufficient understanding of the evolution of crowd behavior in fire scenarios and avoids possible misjudgments at critical moments.

[0066] The multi-level evacuation guidance strategy tree implements a precise guidance strategy that is closely integrated with specific situations. It provides differentiated guidance based on different regions and people with different psychological states, overcoming the problem of traditional guidance strategies being out of touch with actual situations.

[0067] By synchronizing global situation understanding and guidance effect data to the digital twin model, the system can continuously optimize the parameters of each module based on historical data, improve overall system performance, and achieve continuous evolution of the system;

[0068] The application of multi-objective optimization algorithms enables the system to effectively reduce the psychological pressure of the crowd while ensuring evacuation efficiency, avoiding the panic and stampede risks that may be caused by simply pursuing evacuation efficiency while ignoring the psychological factors of the crowd. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a flow chart of a tunnel fire emergency linkage method based on digital twins in the present invention;

[0070] Figure 2 This is a bar chart showing the effect of the context-enhanced attention mechanism on the dynamic adjustment of the weights of different attention heads in a fire scene;

[0071] Figure 3 It is a bar chart analyzing the three main psychological state indicators of people in different regions;

[0072] Figure 4 It is a scatter plot of the distribution of the results of the application of the multi-objective optimization algorithm;

[0073] Figure 5 It is a combination chart for real-time guidance effect evaluation and comparison;

[0074] Figure 6 It is a bar chart comparing the performance of this system and the traditional system in terms of key performance indicators. DETAILED DESCRIPTION

[0075] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0076] At least one embodiment of the present invention discloses a tunnel fire emergency linkage method based on digital twins, such as Figure 1 Shown, including:

[0077] Step 1: Collect multimodal data of the fire environment through various sensors and build a global situation understanding model;

[0078] In this step, the system collects multimodal data of the fire environment through various sensors deployed at different locations in the tunnel (including but not limited to temperature sensors, smoke detectors, infrared cameras, visible light cameras, gas concentration detectors, etc.). Specifically, it includes:

[0079] Step 1.1, data collection;

[0080] Data is collected regularly from each sensor node and transmitted to the central processing system through the communication network in the tunnel;

[0081] Step 1.2, data preprocessing;

[0082] Perform pre-processing operations such as noise reduction, completion, and normalization on the collected data, including:

[0083] Perform minimum and maximum normalization processing on the temperature data and map it to the range of [0, 1];

[0084] The smoke concentration data were standardized and converted into Z scores;

[0085] The gas concentration data is segmented and normalized according to the danger threshold;

[0086] Perform pre-processing on image data such as grayscale conversion and contrast enhancement;

[0087] Step 1.3, spatiotemporal alignment;

[0088] Align data collected by different sensors and locations in time and space to form a unified data representation;

[0089] Step 1.4, feature extraction;

[0090] For data of different modalities, a special feature extraction network is used to extract the corresponding feature representation;

[0091] Step 1.5, multimodal fusion;

[0092] A multimodal fusion network based on the attention mechanism is used to fuse the features of different modalities to construct a global situation representation of the tunnel fire, including the fire spread range, smoke distribution, temperature distribution, and toxic gas concentration distribution.

[0093] The global situation understanding model uses the following structural components:

[0094] Multimodal encoder unit: contains multiple independent encoder components, each encoder is responsible for processing data of a specific modality;

[0095] Cross-modal attention unit: calculates the correlation weight matrix between different modalities;

[0096] Feature fusion unit: performs weighted fusion of each modality feature according to the attention weight;

[0097] Situation representation unit: converts the fused features into a physical quantity distribution representation of the tunnel space.

[0098] Furthermore, the global situation understanding model also includes a time series processing unit to capture the dynamic evolution trend of the fire situation.

[0099] Step 2: Based on the global situation understanding model, a multi-head attention network is used to process the environmental state and crowd behavior characteristics to achieve collaborative perception of environmental and crowd information;

[0100] In this step, the system uses a multi-head attention network to simultaneously process environmental status and crowd behavior characteristics, achieving collaborative perception of environmental and crowd information. The multi-head attention network includes the following structural components:

[0101] Environmental feature input unit: This unit receives and processes environmental state features from the global situation understanding model, such as fire location, smoke concentration, temperature, and oxygen content. This unit also includes an environmental feature preprocessing subunit that standardizes environmental parameters of different physical quantities to ensure that environmental parameters of different dimensions can be effectively compared.

[0102] Crowd feature input unit: receives and processes the behavior characteristics of the crowd, such as position distribution, moving speed, moving direction, and gathering state; this unit includes a crowd feature preprocessing subunit that normalizes position coordinates, converts moving speed into relative speed ratio, and quantifies gathering state into density index, ensuring that different types of behavior characteristics can be represented in the same feature space;

[0103] Multi-head attention calculation unit: contains parallel attention head components, each responsible for focusing on a specific environmental crowd interaction pattern, such as:

[0104] Attention head 1: focuses on the relationship between fire location and crowd distribution;

[0105] Attention head 2: focuses on the relationship between smoke diffusion and crowd moving direction;

[0106] Attention head 3: focuses on the relationship between temperature gradient and crowd moving speed;

[0107] Attention output fusion unit: fuses the outputs of attention heads through learnable weights;

[0108] Feature representation generation unit: generates a comprehensive feature representation that contains both environmental state and crowd behavior information.

[0109] Further, before processing by the multi-head attention network, the system performs feature space alignment preprocessing on environmental features and crowd features, including:

[0110] Feature dimension unification: through feature dimension reduction or dimension increase operations, the environmental features and crowd features have the same dimension;

[0111] Feature scale standardization: Z-score standardization or Min-Max normalization processing is performed on different physical dimension features;

[0112] Feature space mapping: using feature projection matrix, features of different dimensions and different physical meanings are mapped to a common hidden space, so that environmental features (physical quantities) and crowd features (behavior characteristics) can be effectively compared and calculated in the same feature space;

[0113] Feature importance balance: through feature importance scoring, balance weights are applied to environmental features and crowd features to avoid dominance of a certain type of feature in the fusion process.

[0114] Further, the multi-head attention network also includes a residual connection unit to preserve key information in the original input features and prevent information loss during attention calculation.

[0115] Furthermore, each attention head of the multi-head attention network includes three sub-components: query transformation, key transformation, and value transformation, which correspond to the mapping relationship between environmental features and crowd features respectively.

[0116] Step 3: Combined with the characteristics of the fire scene, the context-enhanced attention mechanism is applied to dynamically adjust the attention weight of the crowd feature dimension;

[0117] This step designs a context-enhanced attention mechanism that dynamically adjusts the attention weights for different characteristic dimensions of the crowd based on the characteristics of different fire scenarios. This is one of the core innovations of this invention. The context-enhanced attention mechanism includes the following execution steps:

[0118] Step 3.1, obtaining fire scene characteristic data;

[0119] Fire scene characteristic data includes factors such as fire location, scale, spread speed, and smoke diffusion status;

[0120] Step 3.2, fire scene characteristic feature extraction and coding;

[0121] The fire scene characteristics are extracted and encoded and converted into scene characteristic vector representation, including the following preprocessing operations:

[0122] The spatial coordinates of the fire location features are normalized so that they are within the interval [0, 1];

[0123] Fire size characteristics were converted to a logarithmic scale to handle fire sizes of different magnitudes;

[0124] Standardize the spread speed and smoke diffusion characteristics;

[0125] Discrete scene features (such as fire type, occurrence area type, etc.) are converted into numerical vectors through one-hot encoding;

[0126] Step 3.3, attention weight calculation;

[0127] Based on the scene feature vector, calculate the attention weights for different crowd feature dimensions (such as location, speed, concentration, panic level, etc.);

[0128] Step 3.4, attention weight application;

[0129] Apply the calculated attention weights to the multi-head attention network to dynamically adjust the attention paid to different demographic characteristics.

[0130] Step 3.5, importance scoring;

[0131] Based on the adjusted attention allocation, the importance of crowd behavior characteristics is scored to highlight the most critical features for evacuation guidance in the current scenario.

[0132] Furthermore, before calculating the weights of the context-enhanced attention mechanism, it is necessary to factor in the scene characteristics. and eigenvectors Perform preprocessing to ensure that they can be correlated in the same feature space:

[0133] Scenario characteristic standardization: Standardize the characteristic factors of different scenarios (such as fire location, scale, spread speed, etc.) at the feature level to eliminate dimension and scale differences;

[0134] Feature vector normalization: Perform L2 normalization on the crowd feature vector so that it lies on the unit hypersphere to ensure the consistency of similarity calculation;

[0135] Cross-domain feature mapping: Design specific cross-domain feature transformation functions to map physical scene characteristics and crowd behavior characteristics into a semantically consistent representation space.

[0136] After completing the above preprocessing, the weight calculation formula of the context-enhanced attention mechanism is as follows:

[0137] ;

[0138] in, Indicates the Features The attention weight of Indicates the Scenario-specific factors; represents the summation symbol; Representing scene characteristics For the first Features 's relevance score; Indicates the Importance weight of each scenario characteristic factor; Represents the total number of scene characteristic factors.

[0139] The inner function is the correlation calculation function, defined as:

[0140] ;

[0141] in is a learnable projection matrix used to transform the scene feature vector With the eigenvector Map to the same semantic space for similarity calculation; is the total number of features; represents the transpose symbol; and Respectively represent and feature vectors; represents the natural exponential function; represents the summation symbol; Indicates the Scenario-specific factors; Representing scene characteristics For the first Features 's relevance score.

[0142] Furthermore, the context-enhanced attention mechanism also includes an adaptive threshold adjustment step, which is used to dynamically determine the threshold of feature importance and screen out the truly critical features.

[0143] like Figure 2 The figure below demonstrates the dynamic adjustment of the weights of different attention heads by the context-enhanced attention mechanism in a fire scenario. By comparing the original and adjusted weights, it is clear how the system dynamically adjusts its focus based on the characteristics of the fire, which is one of the core innovations of this patent.

[0144] Step 4: Based on the adjusted attention weights, a psychological-environmental interaction model is constructed to predict the evolution of crowd behavior under different fire scenarios;

[0145] This step builds a psychological-environmental interaction model to predict the evolution of crowd behavior under different fire scenarios. To ensure the effective integration of environmental parameters (physical quantities) and psychological state parameters (subjective indicators), the system first preprocesses these heterogeneous data into a unified representation:

[0146] Unified representation of physical quantities and psychological indicators: Design a dual-domain representation learning framework to map physical environment parameters and psychological state indicators into the same semantic space;

[0147] Multi-scale data standardization: Multi-scale standardization technology is used for data of different temporal and spatial scales to ensure the comparability of data with different sampling frequencies and coverage;

[0148] Heterogeneous data fusion preprocessing: By weighting the importance of features, the influence of environmental physical parameters and psychological state parameters in the model is balanced to prevent a certain type of data from dominating the model behavior due to different numerical ranges or units.

[0149] The mind-environment interaction model includes the following structural components:

[0150] Environmental perception unit: Receives tunnel environment parameters, including smoke density, temperature, visibility, and toxic gas concentrations. This unit normalizes these parameters and calculates the deviation of each parameter from the safety threshold, unifying the danger level of different physical quantities into a danger index in the range [0, 1].

[0151] Psychological state modeling unit: This unit includes a multi-layer perceptron and a state converter, and is used to model psychological states such as the level of panic, herd behavior tendency, and rational decision-making ability of the crowd. This unit converts crowd behavior data into quantitative psychological state indicators, encodes and quantifies qualitative observation data (such as facial expressions and body language), and uniformly maps psychological state information from different sources and types into the same feature space.

[0152] Behavior Evolution Prediction Unit: Based on the sequence prediction network, it predicts the possible behavioral changes of the crowd in the subsequent time period, such as changes in movement direction, speed, aggregation or dispersion trends, etc.

[0153] Dangerous state recognition unit: Identifies possible dangerous states, such as crowd congestion, panic stampede, and lingering in dangerous areas, through pattern matching and threshold judgment;

[0154] Scenario deduction unit: Using the Monte Carlo tree search structure, it deduces multiple possible scenario development paths and evaluates the effectiveness of different guidance strategies.

[0155] Furthermore, the psychological state modeling unit in the psychological environment interaction model adopts a hierarchical structure, including a basic emotion layer, a cognitive evaluation layer, and a behavioral decision layer.

[0156] Furthermore, the psychological-environmental interaction model also includes an individual difference modeling unit, which is used to characterize the differences in responses of individuals from different populations when facing the same environmental threats.

[0157] like Figure 3 The data shows three key psychological indicators for people in different regions: panic level, herd behavior tendency, and rational decision-making ability. This data allows us to intuitively understand the differences in psychological states among people in different regions, providing basic data support for the psychological environment interaction model and helping the system generate more accurate evacuation guidance strategies.

[0158] Step 5: Generate a situation-adaptive multi-level evacuation guidance strategy tree based on the prediction results to provide differentiated evacuation guidance;

[0159] This step generates a situation-adaptive multi-level evacuation guidance strategy tree based on the above analysis results. The multi-level evacuation guidance strategy tree model includes the following structural components:

[0160] Root node unit: represents the starting point of the overall evacuation decision and contains global decision parameters;

[0161] Environmental status branch unit: divides the tunnel space into multiple regional nodes based on the fire situation, such as safe zone nodes, buffer zone nodes, and dangerous zone nodes;

[0162] Psychological state branch unit: Under each regional node, further branches are formed according to the psychological state of the population to form secondary nodes;

[0163] Guidance strategy leaf node unit: The leaf node of the tree contains the specific guidance strategy for a specific combination of environmental state and psychological state, including:

[0164] Visual guidance components: such as dynamic evacuation indicator lights, escape route signs, etc.;

[0165] Voice guidance component: Generates voice prompts with different tones and content for people with different levels of panic;

[0166] Physical guidance components: such as emergency lighting, ventilation equipment controls, etc.;

[0167] Priority Scoring Unit: Assigns a priority score to each leaf node policy to ensure that resources are prioritized for areas and populations that need help the most.

[0168] Furthermore, each leaf node in the multi-level evacuation guidance strategy tree model contains a conditional trigger rule component, which is used to define the specific conditions for strategy activation.

[0169] Furthermore, the multi-level evacuation guidance strategy tree model also includes a strategy conflict resolution unit for handling possible conflicts between multiple strategies.

[0170] In order to generate the optimal evacuation guidance strategy, this step applies a multi-objective optimization algorithm to balance evacuation efficiency and crowd psychological pressure. The multi-objective optimization algorithm includes the following execution steps:

[0171] Define optimization objective functions, including the shortest evacuation time objective function , minimum congestion risk objective function , minimum psychological stress objective function wait;

[0172] Furthermore, these objective functions are defined as follows:

[0173] Evacuation time objective function:

[0174] ;

[0175] in Representation scheme Minimum evacuation time; Assemble for all personnel; For the plan Next The time required for individuals to complete the evacuation; represents the maximum value function; Indicates evacuation plan; Indicates conditional separator; Indicates belonging to a symbol;

[0176] Evacuation time calculation function The specific implementation is: through the evacuation simulation model calculation in the scheme Next The estimated evacuation time for a person from their current location to the nearest safe exit. This function takes into account factors such as the distance from the person's current location to the exit, population density along the route, walking speed, congestion delays, and individual characteristics (such as age and mobility). The calculation uses a grid-based path planning algorithm, dividing the route into multiple grid cells. The travel time is calculated for each cell and the travel time for different sections is accumulated to obtain the total evacuation time.

[0177] Congestion risk objective function:

[0178] ;

[0179] in Representation scheme Minimum congestion risk; Gather for all evacuation routes; For the plan Next Congestion measure for each channel; For the The channel weight of each channel; represents the summation symbol; Indicates belonging to a symbol;

[0180] Congestion measurement function The specific implementation is: calculation in the scheme Lower Passage This function calculates a congestion indicator based on the channel's occupant density and the ratio of traffic flow to capacity. It first determines the channel's standard capacity and then calculates the ratio of the current occupant flow to the standard capacity, while also considering the spatial uniformity of occupants within the channel. When occupant density exceeds a critical threshold, a nonlinear penalty factor is introduced to reflect the rapidly increasing congestion risk. The final output is a normalized congestion index between 0 and 1.

[0181] Psychological stress objective function:

[0182] ;

[0183] in Representation scheme Minimum psychological stress; For the plan Next Individual stress levels; For the plan Next Individuals’ level of uncertainty perception; represents the weight coefficient of stress level, The weight coefficient representing the uncertainty perception level; represents the summation symbol; Indicates the personnel number; Indicates belonging to a symbol; Represents the collection of all personnel;

[0184] Stress level calculation function The specific implementation is: evaluate in the program Next The function measures the psychological stress level of individuals. This function comprehensively considers environmental threat factors (such as fire intensity, temperature, and smoke density), individual factors (such as age, coping skills, and emergency experience), and social factors (such as group behavior, information access, and the clarity of leadership instructions). The calculation process uses a hierarchical weighted approach, first assessing the stress contribution of each sub-factor, then integrating them using adaptive weights to ultimately output a stress level score ranging from 0 to 10.

[0185] Uncertainty perception level calculation function The specific implementation is: measure the Next The uncertainty level of an individual in their current situation. Based on information entropy theory, this function calculates the completeness, accuracy, and consistency of the information a person receives and assesses the predictability of environmental changes. Specific calculations include parameters such as the number of information acquisition channels, information update frequency, information content consistency, and the rate of change of environmental states. Uncertainty perception increases when information is missing, contradictory, or the environment is rapidly changing. Conversely, uncertainty perception decreases when sufficient, consistent, and stable information is available. The final output is an uncertainty score on a scale of 0 to 10.

[0186] Construct a set of constraints, such as the number of available evacuation channels, the number of guidance equipment, and resource allocation constraints;

[0187] Initialize the solution space and generate an initial solution set that meets the constraints;

[0188] Perform an iterative multi-objective optimization process, using Pareto optimization methods to continuously improve the solution set;

[0189] Obtain a Pareto optimal solution set, that is, a set of optimal solutions that balance different objectives;

[0190] The fitness score is calculated based on the current scenario characteristics, and the most suitable evacuation plan is selected from the optimal solution set.

[0191] Further, the fitness score calculation function is defined as:

[0192] ;

[0193] wherein is the fitness score of the solution under the scenario ; is the candidate solution; is the current scenario characteristic; is the dynamic weight of the th objective function under the scenario ; is the normalized result of the th objective function; denotes the summation symbol;

[0194] The objective functions include three objective functions of evacuation time, congestion risk and psychological stress, corresponding to , and , respectively;

[0195] The calculation of the dynamic weight requires preprocessing and standardization, specifically including:

[0196] Scenario characteristic vectorization: converting the scenario characteristic into a numerical feature vector, including fire size, spread speed, smoke density, crowd density and other multi-dimensional features;

[0197] Weight dynamic mapping: designing a scenario-weight mapping function to map the scenario characteristic vector to the weight space, ensuring reasonable weight distribution of the objective functions under different scenarios;

[0198] Weight normalization: normalizing the generated weight vector to ensure that the total weight is 1;

[0199] Weight smoothing mechanism: introducing a weight smoothing factor to avoid drastic fluctuations in weight caused by slight changes in scenarios, ensuring system stability.

[0200] wherein the specific implementation of the normalization function is as follows: different normalization strategies are adopted according to the characteristics of different objective functions, linear normalization is adopted for the evacuation time objective function to map it to the [0, 1] interval, piecewise function normalization is adopted for the congestion risk objective function to reflect the risk threshold effect, and S-shaped function normalization with saturation effect is adopted for the psychological stress objective function to emphasize the marginal effect of high stress state. The function first calculates the quantile of the objective function value in the historical data, then applies the corresponding nonlinear transformation according to the objective characteristics, and finally outputs the normalized value standardized to the [0, 1] interval.

[0201] Furthermore, data preprocessing of different objective function values ​​is required before calculating the fitness score, including:

[0202] The objective function value of evacuation time Min-Max normalization is used to map it to the interval [0, 1];

[0203] Congestion risk objective function value Using piecewise function mapping to reflect the nonlinear growth of risk at different congestion levels;

[0204] Objective function value of psychological stress Perform index conversion and normalization to appropriately emphasize the adverse effects of high stress states;

[0205] Normalization function Different normalization strategies are selected according to the characteristics of different objective functions to ensure that objective functions of different dimensions and scales can be effectively weighted compared.

[0206] Furthermore, the multi-objective optimization algorithm uses a solution sorting step based on dominance relations to improve the efficiency of searching the solution space.

[0207] Furthermore, the multi-objective optimization algorithm adopts an adaptive weight adjustment strategy during the iteration process to adjust the optimization weight of each objective according to the dynamic changes of the scenario.

[0208] like Figure 4 The figure below shows how different candidate solutions generated by the multi-objective optimization algorithm perform in terms of evacuation time and congestion risk. This chart provides an intuitive understanding of the strengths and weaknesses of different solutions and their trade-offs, helping the system find the optimal compromise between multiple objectives. This demonstrates the effectiveness of the multi-objective optimization algorithm employed in this patent.

[0209] Step 6: During the implementation of the evacuation guidance strategy, conduct real-time effect evaluation and optimize the overall system through the digital twin model;

[0210] In this step, the system conducts real-time effectiveness evaluation of the generated evacuation guidance strategy and simultaneously optimizes the entire system through the digital twin model. This step involves two closely related steps: real-time parameter adjustment and digital twin system optimization.

[0211] Step 6.1, real-time guidance effect evaluation and parameter optimization;

[0212] The system evaluates the generated evacuation guidance strategy in real time and optimizes guidance parameters based on feedback. The real-time guidance effect evaluation and parameter optimization algorithm includes the following execution steps:

[0213] Define evaluation indicators for guidance effectiveness, including evacuation speed, smoothness of crowd flow, and panic control;

[0214] Collect real-time data through sensor networks to monitor the actual effects of guiding strategies after execution;

[0215] Compare and calculate the measured effect with the expected effect, analyze the existing deviation, and use the deviation function Calculation; before calculation, standardize and pre-process different types of effect indicator data, including:

[0216] The evacuation speed index (m / s) was normalized and converted into a ratio relative to the maximum possible evacuation speed;

[0217] Normalize the smoothness of crowd flow (dimensionless) so that it is distributed in the interval [0, 1];

[0218] The panic control degree (based on the questionnaire score) was converted into standard scores;

[0219] Apply specific scaling factors to indicators of different units and magnitudes to make their comparison meaningful;

[0220] Furthermore, the deviation function Defined as:

[0221] ;

[0222] in is the deviation value; For the The measured values ​​of the effect indicators (after standardization), For the The expected value of the effect indicator (after the same standardization process), For the The importance weight of each effect indicator, is the total number of effect indicators; represents the summation symbol;

[0223] Based on the deviation analysis results, the parameter adjustment amount is generated to adaptively adjust the guidance strategy parameters;

[0224] Apply the adjusted parameters to the boot device, updating the boot timing, signal strength, boot information content, etc.

[0225] Furthermore, the real-time guidance effect evaluation and parameter optimization algorithm also includes a parameter boundary checking step to ensure that the adjusted parameters are within the valid range.

[0226] Furthermore, the real-time guidance effect evaluation and parameter optimization algorithm adopts an online learning method to gradually optimize the parameter adjustment strategy and improve the adjustment effect.

[0227] As shown in Figure 5 , the key indicators of the system real-time guidance effect evaluation are displayed, the left axis represents the evacuation speed of different areas, and the right axis represents the comprehensive indicator score (the weighted average of flow smoothness and panic control). By comparing the indicator performance of different areas, the effectiveness of the guidance strategy can be evaluated, providing a basis for parameter optimization, and reflecting the role of real-time guidance effect evaluation and parameter optimization mechanism in this patent.

[0228] Step 6.2, digital twin model synchronization and system optimization;

[0229] This link synchronizes global situation understanding and guidance effect data to the digital twin model, supports overall system optimization, and forms a closed-loop feedback mechanism of real-time adjustment and long-term optimization. The digital twin model includes the following structural components:

[0230] Data synchronization unit: responsible for receiving and processing real-time acquired fire situation data, crowd behavior data, and guidance effect data;

[0231] Physical-virtual data conversion preprocessing unit: necessary data preprocessing and format conversion before mapping physical world data to virtual model, including:

[0232] Data format unification: convert physical world data of different sources and formats into standard data format acceptable by the digital twin model;

[0233] Space-time reference system conversion: establish the coordinate mapping relationship between physical space and virtual space to ensure accurate spatial position data correspondence;

[0234] Data precision matching: according to the precision requirements of the virtual model, perform upsampling or downsampling processing on the physical world data;

[0235] Abnormal value detection and processing: identify and process abnormal values, missing values and noise in physical sensor data to ensure data quality;

[0236] Data semantic enhancement: add semantic labels and context information to the original data to improve the understanding ability of the virtual model to the data.

[0237] Virtual mapping unit: map the preprocessed physical entity information to the corresponding entity in the virtual space;

[0238] Three-dimensional visualization unit: visually display fire situation, crowd distribution, evacuation path, etc. through graphics rendering technology;

[0239] Historical data storage unit: save the historical data of system operation in time series form;

[0240] Parameter optimization unit: Based on historical data analysis, it proposes parameter optimization suggestions for each module of the system;

[0241] Simulation verification unit: Verify the performance of the optimized system through simulation experiments in digital space.

[0242] Furthermore, the digital twin model also includes a state prediction unit, which can predict the future evolution trend of the system based on the current state.

[0243] Furthermore, the digital twin model includes a physical-virtual bidirectional interaction unit, which supports bidirectional information flow and control between the physical world and the virtual world.

[0244] like Figure 6 The figure shows the differences between this system and traditional systems in key performance indicators, including collaborative perception capability and evacuation guidance effectiveness. Through intuitive comparison, it can be seen that this system outperforms traditional systems in all indicators, verifying the effectiveness and advancement of the core technical solution in this patent.

[0245] A tunnel fire emergency linkage system based on digital twins, used to implement the above-mentioned tunnel fire emergency linkage method based on digital twins, comprising:

[0246] Situational awareness module, used to obtain fire environment data through sensor networks and build an overall situation;

[0247] Collaborative analysis module, used to process environmental and crowd characteristic information and achieve multi-dimensional information fusion;

[0248] Adaptive attention module, used to dynamically assign attention priorities based on fire development characteristics;

[0249] Behavior prediction module, used to build psychological response and behavior models of people under different environmental conditions;

[0250] Strategy generation module, used to create differentiated evacuation guidance plans based on scene characteristics and crowd status;

[0251] The system optimization module is used to evaluate the guidance effect and realize dynamic parameter adjustment through virtual-real mapping.

[0252] Here, this embodiment provides an application example:

[0253] The application scenario of the present embodiment is a certain city's two-way six-lane urban tunnel, with a total length of 3.5 kilometers and a single-way three-lane, with an average daily traffic volume of about 80,000 vehicles. The tunnel is equipped with a complete sensor network, including temperature sensors (140 groups), smoke detectors (280 groups), infrared and visible light cameras (210 groups), toxic gas detectors (70 groups), LED dynamic evacuation indicator lights (700 groups), and a voice broadcasting system (70 groups). On June 15, 2023, a chemical truck collision and fire accident occurred in the tunnel, and there were about 420 people in the tunnel who needed to be evacuated urgently. The system was applied in this emergency and the technical effect was verified.

[0254] After the accident, various sensors in the tunnel quickly collected environmental data and pre-processed them (temperature data normalization, smoke concentration standardization, image data feature extraction, etc.). The fire situation assessment results generated by the system after processing are shown in Table 1:

[0255] Table 1: Fire situation assessment results

[0256]

[0257] At the same time, the system collects crowd distribution and behavior data through cameras and sensors, and identifies and analyzes the crowd state in different areas. The multi-head attention network processes both environmental state and crowd behavior features, focusing on different environmental-crowd interaction patterns. The different regional crowd states identified by the system are shown in Table 2:

[0258] Table 2: Different regional crowd state identification results

[0259]

[0260] The context-enhanced attention mechanism dynamically adjusts the attention weights of different feature dimensions of the crowd according to the current fire scene characteristics. The system identifies that the fire spreads rapidly, the smoke concentration is extremely high, and the crowd is in panic and some reverse movement, so it automatically increases the attention weights of smoke diffusion-movement direction and toxic gas concentration-crowd density, ensuring that these key factors are focused on, and the originally evenly distributed weights (0.25 each) are dynamically adjusted to a more suitable distribution for the current scene (smoke diffusion-movement direction: 0.35, toxic gas concentration-crowd density: 0.30, fire location-crowd distribution: 0.20, temperature gradient-movement speed: 0.15).

[0261] The psychological environment interaction model predicts the evolution of crowd behavior in the fire scene based on environmental state and crowd behavior data. The system models the psychological state and risk assessment of different areas, and the psychological state modeling and evacuation guidance strategy are shown in Table 3:

[0262] Table 3: Psychological state modeling and evacuation guidance strategy

[0263]

[0264] Based on fire situation assessments and the psychological state of the crowd, the system constructed a multi-level evacuation guidance strategy tree. This strategy tree uses environmental and psychological conditions as branching conditions, generating highly targeted guidance strategies. The system assigns guidance priorities to different areas, ensuring that resources are prioritized for those most at risk and those most in need. For example, the 0-100 meter area (the extremely high panic zone) uses brief, clear instructions and a flashing red arrow at a 2Hz frequency, while also reducing the alarm volume to avoid increasing panic. In contrast, more conventional guidance methods are used for areas further away from the incident.

[0265] During the evacuation process, the system evaluates the effectiveness of guidance strategies in real time and optimizes guidance parameters based on feedback. By collecting real-time data through a sensor network, the system continuously monitors key indicators such as evacuation speed, crowd flow smoothness, and panic control. It uses a deviation function to calculate the deviation between actual and expected results and generates appropriate parameter adjustment plans. The system's evaluation and adjustment results are shown in Table 4.

[0266] Table 4: Guidance strategy effect evaluation and parameter adjustment

[0267]

[0268] During the evacuation process, the system synchronized global situational understanding and guidance effect data to the digital twin model, supporting overall system optimization. The digital twin model constructed a virtual mapping of the physical tunnel, enabling two-way interaction between the physical and virtual worlds. Through this model, the system was able to map physical world data into virtual space for visualization, future state prediction, and parameter optimization. The prediction and optimization results of the digital twin model at key time points are shown in Table 5:

[0269] Table 5: Digital twin model prediction and optimization results

[0270]

[0271] During the evacuation process, the system applied a multi-objective optimization algorithm to balance evacuation efficiency with the psychological stress of the crowd. This algorithm simultaneously optimized three key objectives: shortest evacuation time, minimum congestion risk, and minimum psychological stress, taking into account constraints such as the number of channels and guidance equipment resources. The system used Pareto optimization to generate multiple candidate solutions and calculated fitness scores based on the current scenario characteristics to select the most suitable evacuation solution. The evaluation results of the candidate solutions generated by the system are shown in Table 6:

[0272] Table 6: Evaluation results of multi-objective optimization schemes

[0273]

[0274] The system selected Plan A as the final implementation plan. Although its evacuation time is slightly longer than Plan B, after comprehensively considering the congestion risk and psychological pressure, its overall adaptability score is the highest. It can effectively reduce the psychological pressure of the crowd while ensuring evacuation efficiency, avoiding the congestion and panic risks that may be caused by simply pursuing evacuation time while ignoring other factors.

[0275] By comparing the performance of this technical solution with that of traditional tunnel fire emergency systems under the same conditions, two core technical effects were verified: the ability to collaboratively perceive and analyze the fire situation and the psychological state of the crowd; and the differentiated and precise evacuation guidance effect.

[0276] The comprehensive comparison results of system performance are shown in Table 7:

[0277] Table 7: Comprehensive comparison of system performance

[0278]

[0279] In terms of collaborative perception capabilities, the system improved the recognition rate of the correlation between environmental factors and crowd behavior and increased the accuracy of psychological state assessment by 41.3%, enabling the system to more accurately understand complex fire scenarios. In terms of evacuation guidance effectiveness, the system shortened evacuation completion time by 31.1%, while also reducing the incidence of local congestion and the proportion of people traveling against traffic, and improving the responsiveness of guidance strategies.

[0280] The system's core innovation, the context-enhanced attention mechanism, effectively bridges the gap between understanding fire situations and analyzing crowd psychology, providing a solid foundation for precise guidance. By dynamically adjusting the weight of attention given to different environmental and crowd characteristics, the system can develop targeted evacuation strategies, improving evacuation efficiency and safety while reducing potential casualties in accidents.

[0281] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of equivalent embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A tunnel fire emergency linkage method based on digital twin, characterized in that: include: Collect multimodal data of the fire environment through various sensors and build a global situation understanding model; Based on the global situation understanding model, a multi-head attention network is used to process environmental status and crowd behavior characteristics to achieve collaborative perception of environmental and crowd information; Combined with the characteristics of fire scenes, the context-enhanced attention mechanism is applied to dynamically adjust the attention weight of crowd feature dimensions; Based on the adjusted attention weights, a mind-environment interaction model is constructed to predict the evolution of crowd behavior under different fire scenarios. Based on the prediction results, a situation-adaptive multi-level evacuation guidance strategy tree is generated to provide differentiated evacuation guidance; During the implementation of the evacuation guidance strategy, real-time effect evaluation is carried out, and the overall system optimization is achieved through the digital twin model.

2. A tunnel fire emergency linkage method based on digital twin according to claim 1, characterized in that: The global situation understanding model includes: A multimodal encoder unit for processing sensor data of different modalities; Cross-modal attention unit, used to calculate the correlation weight matrix between different modalities; Feature fusion unit, used to perform weighted fusion of each modality feature according to the attention weight; The situation representation unit is used to convert the fused features into a physical quantity distribution representation of the tunnel space.

3. A tunnel fire emergency linkage method based on digital twin according to claim 1, characterized in that: The multi-head attention network includes: Environmental feature input unit, used to receive and process environmental state features in the global situation understanding model; Crowd feature input unit, used to receive and process crowd behavior features; Multi-head attention computing unit, including Each attention head is responsible for focusing on different interaction patterns between people in the environment; Attention output fusion unit is used to The outputs of the attention heads are weighted fused.

4. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The psychological environment interaction model includes: An environment sensing unit, configured to receive environment state parameters; Psychological state modeling unit, used to model the psychological state of the crowd; Behavior evolution prediction unit, used to predict changes in crowd behavior; The dangerous state identification unit is used to identify the dangerous state.

5. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The multi-level evacuation guidance strategy tree includes: The root node unit represents the starting point of the overall evacuation decision; Environmental state branch unit, used to divide the tunnel space into different area nodes; Psychological state branching unit, used for further branching based on the psychological state of the crowd; The guidance strategy leaf node unit contains the specific guidance strategy for a specific combination of environmental state and psychological state; The priority scoring unit is used to assign a priority to each guidance strategy.

6. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The context-enhanced attention mechanism includes: Obtain fire scene characteristic data, including fire location, scale, spread speed, and smoke diffusion status; Extract and encode the fire scene characteristics and convert them into scene characteristic vector representation; According to the scene feature vector, calculate the attention weights for different crowd feature dimensions; Apply the calculated attention weights to the multi-head attention network to dynamically adjust the attention paid to different demographic characteristics. Based on the adjusted attention allocation, the importance of crowd behavior characteristics is scored to highlight the most critical features for evacuation guidance in the current scenario.

7. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The real-time effect evaluation includes: Define evaluation indicators for guidance effectiveness, including evacuation speed, smoothness of crowd flow, and panic control; Collect real-time data through sensor networks to monitor the effectiveness of guiding strategy execution; The deviation function is used to calculate the deviation between the measured effect and the expected effect; Generate parameter adjustments based on the deviation analysis results.

8. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The digital twin model includes: A data synchronization unit, used to receive and process the situation data acquired in real time; A virtual mapping unit, configured to map physical entity information to a virtual space; 3D visualization unit, used to display fire situation and crowd distribution; Historical data storage unit, used to save historical data of system operation; Parameter optimization unit, used to optimize the parameters of each module of the system.

9. The tunnel fire emergency linkage method based on digital twin according to claim 1 is characterized in that: The generation of the scenario-adaptive multi-level evacuation guidance strategy tree uses a multi-objective optimization algorithm, which includes: Define three optimization goals: shortest evacuation time, minimum congestion risk, and lowest psychological stress; Construct a set of constraints, including the number of evacuation channels and the number of guidance equipment; Use Pareto optimization method to solve multi-objective optimization problems; According to the characteristics of the scenario, the most suitable evacuation plan is selected from the optimal solution set.

10. A tunnel fire emergency linkage system based on digital twins, characterized in that: A tunnel fire emergency linkage method based on digital twins for executing any one of claims 1 to 9, comprising: Situational awareness module, used to obtain fire environment data through sensor networks and build an overall situation; Collaborative analysis module, used to process environmental and crowd characteristic information and achieve multi-dimensional information fusion; Adaptive attention module, used to dynamically assign attention priorities based on fire development characteristics; Behavior prediction module, used to build psychological response and behavior models of people under different environmental conditions; Strategy generation module, used to create differentiated evacuation guidance plans based on scene characteristics and crowd status; The system optimization module is used to evaluate the guidance effect and realize dynamic parameter adjustment through virtual-real mapping.

Images