Vehicle fire escape path generation method, device and electronic equipment

Abstract

The application provides a vehicle fire refuge path generation method and device and electronic equipment, relates to the field of automobile intelligent driving and active safety technology, and comprises the following steps: acquiring thermal imaging data and gas concentration data of the surrounding environment of a vehicle, determining a fire source position based on the thermal imaging data and the gas concentration data; calculating the radiant heat flux value of each space point in the environmental space through a preset physical diffusion model according to the fire source position and environmental wind field parameters, and mapping the radiant heat flux value to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information; determining a path search generation value based on the thermal hazard level information of each grid unit in the dynamic thermal grid map, and generating a refuge path leading to a safe area through the path search generation value. The application upgrades from static geometric obstacle avoidance to dynamic thermodynamic safety path planning, and effectively avoids secondary thermal injury in the refuge process.

Description

Technical Field

[0001] This application relates to the field of intelligent driving and active safety technology for automobiles, and in particular to a method, device and electronic device for generating vehicle fire avoidance paths. Background Technology

[0002] With the popularization of new energy vehicles, existing technologies for proactive risk avoidance when vehicles face external fire risks in scenarios such as parking lots and tunnels typically use radar or visual sensors to detect fire sources and treat them as static obstacles for geometric obstacle avoidance planning. However, in real fires, the height of flames, the range of heat radiation, and the direction of smoke diffusion change dynamically over time. Relying solely on geometric distance planning may cause vehicles to enter high-temperature smoke recirculation zones or areas of strong heat radiation, leading to "pseudo-safety" risks such as tire blowouts or high-pressure system failures. Summary of the Invention

[0003] The purpose of this application is to provide a method, device, and electronic device for generating vehicle fire escape routes, so as to alleviate the above-mentioned technical problems existing in the prior art.

[0004] In a first aspect, the present invention provides a method for generating vehicle fire escape routes, comprising: Acquire thermal imaging data and gas concentration data of the environment surrounding the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data; Based on the fire source location and environmental wind field parameters, the radiative heat flux value of each spatial point in the environmental space is calculated through a preset physical diffusion model, and the radiative heat flux value is mapped to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information. Based on the thermal hazard level information of each grid cell in the dynamic thermal grid, the path search cost is determined, and an escape path to a safe area is generated using the path search cost.

[0005] In an optional implementation, determining the location of the fire source based on the thermal imaging data and gas concentration data includes: The thermal imaging data is processed for image segmentation to identify high-temperature areas; Calculate the concentration gradient direction of the gas concentration data; The location of the high-temperature region is corrected based on the direction of the concentration gradient to determine the location of the fire source.

[0006] In an optional implementation, based on the fire source location and environmental wind field parameters, the radiative heat flux value of each spatial point in the environmental space is calculated using a preset physical diffusion model, and the radiative heat flux value is mapped to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information, including: Based on the location of the fire source and the environmental wind field parameters, the radiative heat flux and smoke concentration values ​​at each spatial point in the environment are calculated using a preset physical diffusion model. The radiant heat flux value is compared with a preset vehicle component thermal failure threshold, and the flue gas concentration value is compared with a preset environmental safety threshold. Based on the comparison results, the thermal hazard level of each spatial point is determined, and the spatial points with the thermal hazard level are mapped to the rasterized space to generate the dynamic thermal raster map.

[0007] In an optional implementation, based on the location of the fire source and the environmental wind field parameters, the radiative heat flux and smoke concentration at each point in the environmental space are calculated using a preset physical diffusion model, including: Obtain the height parameters of the confined space in the vehicle's environment; The Froude number is calculated based on the environmental wind field parameters and the confined space height parameters. The preset physical diffusion model is modified based on the fire source location and the Froude number to calculate the radiative heat flux and smoke concentration at each spatial point in the environment.

[0008] In an optional implementation, the path search cost is determined based on the thermal hazard level information of each grid cell in the dynamic thermal grid diagram, and an escape path to a safe area is generated using the path search cost, including: The distance cost is determined based on the geometric distance between nodes during the path search process, the thermal hazard cost is determined based on the thermal hazard level information in each grid cell of the dynamic thermal grid map, and the dynamic threat cost is determined based on the urgency of the fire. The first weighting coefficient corresponding to the thermal hazard cost and the second weighting coefficient corresponding to the dynamic threat cost are determined based on the current fire situation. The distance cost, the thermal hazard cost under the first weighting coefficient, and the dynamic threat cost under the second weighting coefficient are input into a preset comprehensive cost function for calculation. By minimizing the total value, a risk avoidance path to the safe zone is generated.

[0009] In an optional implementation, determining a first weighting coefficient corresponding to the thermal hazard cost and a second weighting coefficient corresponding to the dynamic threat cost based on the current fire situation includes: When a local open flame is detected and the ambient wind speed is less than a preset wind speed threshold, it is determined to be a medium-risk mode, and the first weight coefficient and the second weight coefficient corresponding to the medium-risk mode are determined respectively. When a large-area open flame is detected or the ambient wind speed is greater than or equal to the preset wind speed threshold, it is determined to be a high-risk mode, and the first weight coefficient and the second weight coefficient corresponding to the high-risk mode are determined respectively. In the high-risk mode, both the first weighting coefficient and the second weighting coefficient are greater than those in the medium-risk mode.

[0010] In an optional implementation, after generating an evacuation path to a safe zone using the path search cost, the method further includes: Real-time monitoring of changes in ambient wind direction and heat release rate from fire sources; When the detected wind direction change angle exceeds a preset angle threshold or the heat release rate of the fire source increases beyond a preset proportion threshold, the evacuation path is dynamically replanned.

[0011] In an optional implementation, after generating an evacuation path to a safe zone using the path search cost, the method further includes: Detect the occupant status inside the vehicle; If no occupants are detected inside the vehicle, the vehicle will be controlled to automatically travel to the safe area according to the described avoidance path. If an occupant is detected inside the vehicle, the dynamic thermal grid map and the hazard avoidance path are projected onto the windshield via an augmented reality head-up display device, assisting the driver in following the hazard avoidance path.

[0012] Secondly, the present invention provides a vehicle fire escape route generation device, comprising: The fire source location module is used to acquire thermal imaging data and gas concentration data of the environment around the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data. The thermal field modeling module is used to calculate the radiative heat flux value of each spatial point in the environment space according to the fire source location and environmental wind field parameters through a preset physical diffusion model, and to map the radiative heat flux value to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information. The path planning module is used to determine the path search cost based on the thermal hazard level information of each grid cell in the dynamic thermal grid map, and to generate an avoidance path to a safe area based on the path search cost.

[0013] Thirdly, the present invention provides a vehicle fire escape route generation system, comprising: The hazard perception module is used to collect thermal imaging data and gas concentration data of the vehicle's surrounding environment; The central processing and decision-making unit is connected to the hazard perception module and is used to receive the thermal imaging data and gas concentration data, and execute the vehicle fire evacuation path generation method as described in any of the foregoing embodiments to generate an evacuation path to a safe area. The vehicle execution control module, connected to the central processing and decision-making unit, is used to receive the avoidance path and control the vehicle to travel according to the avoidance path.

[0014] Fourthly, the present invention provides an electronic device including a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method described in any of the foregoing embodiments.

[0015] Fifthly, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when invoked and executed by a processor, cause the processor to perform the method described in any of the foregoing embodiments.

[0016] The vehicle fire escape path generation method, device, and electronic equipment provided in this application determine the fire source location by acquiring thermal imaging data and gas concentration data and cross-validating them. Thermal imaging identifies high-temperature areas, and gas concentration gradients correct for location drift caused by airflow interference, solving the problem of inaccurate fire source location under smoke obstruction or airflow interference with a single sensor. This provides accurate boundary conditions for subsequent thermal field modeling. Based on the fire source location and environmental wind field parameters, radiative heat flux is calculated using a physical diffusion model to generate a dynamic thermal grid map. The dynamic changes in thermal radiation and smoke diffusion are incorporated into the environmental model, enabling path planning to perceive the fire spread trend rather than relying solely on a fixed fire source location, thus overcoming the limitation of static obstacle avoidance logic in handling dynamic fires. Thermal hazard level information is used as part of the path search cost to generate escape paths, allowing path planning to consider thermal safety while achieving geometric obstacle avoidance. The output path not only avoids physical obstacles but also avoids high-temperature radiation and smoke areas, solving the problem of geometrically safe but actually high-temperature pseudo-safe paths. Overall, this application upgrades static geometric obstacle avoidance to dynamic thermodynamic safety path planning, effectively preventing secondary thermal injuries during the escape process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a method for generating vehicle fire escape routes, provided in this application embodiment; Figure 2 A flowchart illustrating the complete control logic from system power-on to disaster avoidance completion, provided in this application embodiment; Figure 3 A path planning flowchart for integrating thermal cost functions is provided in an embodiment of this application; Figure 4 A structural diagram of a vehicle fire escape path generation device provided in an embodiment of this application; Figure 5 A structural diagram of a vehicle fire escape path generation system provided in this application embodiment; Figure 6 This is a structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] This application provides a method for generating vehicle fire escape paths. This method, when a vehicle encounters an external fire, establishes a dynamic model of the thermal radiation field by fusing multimodal sensing data to generate an escape path that considers thermal hazards. When a vehicle encounters an external fire in scenarios such as parking lots, tunnels, and forest areas, traditional geometric obstacle avoidance path planning only considers the distance to obstacles, neglecting the dynamic changes in physical fields such as thermal radiation and smoke diffusion. This may result in the vehicle entering a high-temperature zone even though it is geometrically far from the fire source. This method constructs a dynamic thermal grid map and embeds the thermal hazard cost into the path search process, ensuring that the generated escape path is both geometrically feasible and thermally safe. See also Figure 1As shown, the method mainly includes the following steps.

[0023] S110, acquire thermal imaging data and gas concentration data of the environment around the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data.

[0024] Thermal imaging data refers to temperature distribution images of the environment surrounding a vehicle, acquired by an infrared thermal imaging camera. This data reflects the distribution of high-temperature areas in the environment. Gas concentration data refers to the concentration values ​​of carbon monoxide and carbon dioxide in the environment surrounding a vehicle, acquired by aerosol sensors (including CO and CO2 sensors). The fire source location refers to the three-dimensional spatial coordinates of the center of the heat source that caused the fire, within the vehicle's coordinate system.

[0025] In practice, infrared thermal imaging cameras positioned on the vehicle's front bumper, roof, or rearview mirrors continuously acquire thermal images of the vehicle's surroundings, while aerosol sensors deployed around the vehicle continuously monitor CO and CO2 concentrations. When the gas concentration exceeds a preset background threshold, the system is activated. The infrared thermal imaging cameras start and perform image segmentation processing on the acquired thermal images (e.g., using a lightweight convolutional neural network) to segment high-temperature regions from the temperature distribution and identify the outline and center point of the flame core area. Simultaneously, the system calculates the concentration gradient direction based on the concentration values ​​from multiple aerosol sensors—the direction of the fastest concentration increase points towards the fire source. However, because ambient airflow (wind) can affect the positioning accuracy of the gas sensors, the gas concentration gradient direction may drift. Therefore, the system uses a particle filtering algorithm to fuse the high-temperature region location obtained from the thermal image segmentation with the gas concentration gradient direction. Using the high-temperature region identified by thermal imaging as the primary positioning information and the gas concentration gradient direction as a correction factor, the system corrects the high-temperature region's location, ultimately outputting accurate fire source coordinates.

[0026] S120, based on the fire source location and environmental wind field parameters, calculate the radiative heat flux value of each spatial point in the environmental space through a preset physical diffusion model, and map the radiative heat flux value to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information.

[0027] Radiant heat flux refers to the thermal radiation power received per unit area, measured in watts per square meter (W / m²), reflecting the intensity of thermal radiation received at that location. Environmental wind field parameters, including wind speed and direction, are obtained through vehicle airflow sensors or meteorological service APIs. A dynamic thermal grid map is a two-dimensional or three-dimensional map that divides the space surrounding the vehicle into regular grid cells, each assigned a thermal hazard level label.

[0028] In practical implementation, the system acquires parameters such as current ambient wind speed, wind direction, air humidity, and ambient temperature, and uses the location of the fire source as input. Then, based on the Gaussian smoke diffusion model and the point source thermal radiation model, it calculates the radiative heat flux and smoke concentration at each spatial point in the environment. The formula for calculating the radiative heat flux q''(x,y) is q''(x,y) = τ·Q·η / (4πr²), where Q is the heat release rate of the fire source (estimated based on the flame area), τ is the atmospheric transmittance, η is the viewing angle coefficient, and r is the distance from the spatial point to the center of the fire source. The smoke concentration is calculated based on the Gaussian diffusion model, considering the influence of wind speed and atmospheric stability on diffusion.

[0029] After calculating the radiative heat flux and flue gas concentration at each spatial point, the system compares the radiative heat flux with preset vehicle component thermal failure thresholds and the flue gas concentration with preset environmental safety thresholds. The vehicle component thermal failure thresholds include at least tire thermal failure thresholds and high-voltage electrical system thermal failure thresholds—tires are at risk of bursting when exposed to heat for extended periods, and high-voltage wiring harnesses and connectors may experience insulation failure at high temperatures. The flue gas concentration environmental safety thresholds refer to relevant standards (e.g., GB / T 18664-2002), with CO concentrations reaching 1700 ppm or CO2 concentrations reaching 4% considered lethal. Based on the comparison results, the system classifies each spatial point into one of four thermal hazard levels: Level 0 (Safe) – radiant heat flux below 200 W / m²; Level 1 (Warning) – radiant heat flux between 200 W / m² and 500 W / m²; Level 2 (Hazardous) – radiant heat flux between 500 W / m² and 800 W / m²; Level 3 (Lethal) – radiant heat flux above 800 W / m² or smoke concentration exceeding the lethal threshold. Subsequently, the system maps the thermal hazard level of each spatial point to a rasterized space, generating a dynamic thermal raster map containing thermal hazard level information. This thermal raster map is overlaid with a geometric obstacle raster map to form a complete environmental model.

[0030] S130, determine the path search cost based on the thermal hazard level information of each grid unit in the dynamic thermal grid diagram, and generate an avoidance path to a safe area through the path search cost.

[0031] A hazard avoidance route refers to the driving trajectory from the vehicle's current location, bypassing the fire source and high-temperature area, and ultimately reaching a safe area. A safe area is a geometrically passable location that is sufficiently far from the fire source (radiant heat flux below 200W / m², thermal hazard level 0).

[0032] In practice, the path planner (e.g., the improved A* algorithm) reads the thermal hazard level information of each grid cell in the dynamic thermal grid during the search process. The total cost of the path search consists of three parts: distance cost (the actual travel distance from the starting point to the current node), thermal hazard cost (determined based on the thermal hazard level of the grid where the current node is located), and dynamic threat cost (the probability of fire expansion when a vehicle arrives at the point, predicted based on the fire spread rate). Specifically, the thermal hazard cost is determined as follows: if the grid's thermal hazard level is 3, the grid is marked as impassable, and the path search algorithm will not include it in the candidate nodes; if the level is 2, the thermal hazard cost is set to the highest value (e.g., 1000); if the level is 1, the thermal hazard cost is set to the second medium value (e.g., 50); if the level is 0, the thermal hazard cost is 0. The dynamic threat cost calculates the probability of fire expansion when a vehicle arrives at the point based on the fire spread rate predicted by parameters such as the fire source heat release rate and wind speed. During the search process, the algorithm compares the total cost of each candidate path and selects the path with the lowest total cost and all nodes being passable as the final avoidance path. Even if the path with the shortest geometric distance passes through a high-temperature area, it will be abandoned due to its high thermal hazard cost, and the algorithm will choose an alternative path, thus achieving path planning that prioritizes thermal safety.

[0033] The vehicle fire escape path generation method provided in this application determines the fire source location by acquiring and fusing thermal imaging data and gas concentration data, avoiding the location drift problem of a single sensor in complex environments. Based on the fire source location and environmental wind field parameters, it calculates radiative heat flux using a physical diffusion model and generates a dynamic thermal grid map, enabling the environmental model to reflect the dynamic changes in heat radiation and smoke diffusion, rather than relying solely on fixed geometric distances. Based on thermal hazard level information, it determines the path search cost and generates escape paths, ensuring that path planning considers both geometric feasibility and thermal safety. The generated paths geometrically avoid obstacles and thermodynamically avoid high-temperature and smoke-filled areas. This method solves the problem in existing technologies where path planning is based solely on geometric obstacle avoidance and fails to consider the potential secondary damage caused by dynamic changes in heat radiation.

[0034] For ease of understanding, the method provided in this application will be described in detail below.

[0035] In one specific implementation, the method for determining the location of the fire source based on thermal imaging data and gas concentration data is as follows.

[0036] Thermal images acquired by an infrared thermal imaging camera are input into a lightweight convolutional neural network (e.g., a semantic segmentation network with a U-Net structure). This network is pre-trained on a training dataset labeled with fire source regions and is capable of pixel-level segmentation of the input thermal images, separating the contours of high-temperature regions from the entire thermal image and calculating the geometric center of these regions. The center position of the high-temperature region provides the approximate location of the fire source, but due to potential biases in thermal imaging under smoke obscuration conditions, gas concentration data is needed for supplementary correction.

[0037] Simultaneously, aerosol sensors deployed at multiple locations around the vehicle continuously collect CO and CO2 concentration values. The system calculates the concentration gradient direction based on the concentration values ​​from multiple sensors—the direction of the fastest concentration increase points towards the fire source. Specifically, the system compares the concentration differences between sensors at different locations to establish a concentration gradient vector field. However, ambient airflow can cause the gas diffusion direction to deviate from the fire source direction, resulting in a drift in the concentration gradient direction. Therefore, the system inputs the location of the high-temperature region obtained from thermal imaging segmentation and the gas concentration gradient direction into a particle filter for fusion. The particle filter uses the location of the high-temperature region identified by thermal imaging as the observation value and the gas concentration gradient direction as the correction value. Through particle resampling and weight updates, it gradually converges to the optimal estimated location, which is the final fire source location. Through this cross-validation method, thermal imaging data provides information on the shape and center location of the high-temperature region, while gas concentration data provides information on the diffusion direction of combustion products. The two complement each other, thereby improving the accuracy of fire source localization.

[0038] In one specific implementation, the above-mentioned method for calculating the radiative heat flux value and generating a dynamic thermal grid map based on the fire source location and environmental wind field parameters using a preset physical diffusion model is as follows.

[0039] First, obtain parameters such as ambient wind speed (unit: m / s), wind direction (azimuth angle relative to the vehicle coordinate system), and air temperature. Then, using the previously determined fire source location as the center, calculate the radiative heat flux and smoke concentration at each spatial point in the environment based on the Gaussian smoke diffusion model and the point source thermal radiation model. The formula for calculating the radiative heat flux q''(x,y) is: q''(x,y) = τ·Q·η / (4πr²) Where Q is the heat release rate of the fire source (in W, estimated from the flame area in the thermal imaging image), τ is the atmospheric transmittance (corrected for ambient humidity), η is the viewing angle coefficient (calculated based on the relative position of the spatial point to the center of the fire source), and r is the distance from the spatial point to the center of the fire source (in meters). The smoke concentration is calculated using a Gaussian diffusion model, taking into account the influence of wind speed on the diffusion range and direction.

[0040] After calculation, the radiative heat flux values ​​at each spatial point are compared with preset thermal failure thresholds for vehicle components. These thresholds include tire thermal failure thresholds and high-voltage electrical system thermal failure thresholds—tires exposed to heat radiation exceeding 500 W / m² for extended periods risk bursting, and high-voltage wiring harnesses and connectors may experience insulation failure at levels exceeding 800 W / m². Simultaneously, the flue gas concentration values ​​are compared with preset environmental safety thresholds (e.g., CO concentration of 1700 ppm or CO2 concentration of 4%).

[0041] In one example, based on the comparison results, spatial points with radiant heat flux below 200 W / m² and smoke concentration below the threshold can be classified as Level 0 (safe), those between 200 and 500 W / m² as Level 1 (warning), those between 500 and 800 W / m² as Level 2 (dangerous), and those above 800 W / m² or with smoke concentration exceeding the threshold as Level 3 (lethal). Each spatial point and its thermal hazard level is then mapped to a rasterized space, generating a dynamic thermal raster map containing thermal hazard level information. This thermal raster map is updated at regular time intervals (e.g., 1 second) to reflect fire spread and environmental changes in real time.

[0042] In one specific implementation, for confined spaces such as underground garages or tunnels, the calculation of radiative heat flux and flue gas concentration values ​​using the preset physical diffusion model needs to be corrected.

[0043] Obtain the confined space height parameter H (in meters) of the vehicle's environment, such as the floor height of an underground parking garage or the height of a tunnel. Then, calculate the Froude number based on the environmental wind field parameter (wind speed v_wind) and the confined space height parameter H: Fr = v_wind / sqrt(g·H), where g is the acceleration due to gravity. The Froude number reflects the relative intensity of buoyancy-driven flow and inertial-driven flow, and is used to determine the flow pattern of flue gas within a confined space.

[0044] Based on the Froude number calculations, the preset physical diffusion model is modified. When Fr is less than a preset threshold (e.g., Fr < 0.5), it indicates that the hot flue gas layer will stabilize below the ceiling, forming a stable stratified flow. In this case, the system adjusts the vertical diffusion coefficient in the flue gas diffusion model so that the diffusion rate of the flue gas along the ceiling direction is higher than the diffusion rate in the downward direction. When Fr is greater than or equal to this threshold, it indicates that inertial forces dominate, and the flue gas mixes more thoroughly with the air; in this case, the system uses a neutral diffusion model. By introducing Froude number corrections, the calculated results of radiative heat flux and flue gas concentration more closely resemble the real physical environment of confined spaces such as underground parking garages or tunnels.

[0045] In one alternative implementation, to upgrade path planning from reactive decision-making to "avoiding future dangers," this application also provides a digital twin-based fire evolution prediction and forward-looking evacuation path generation scheme. Specifically, the system constructs a digital twin model of the fire scene around the vehicle on an onboard computing platform. This digital twin model includes a lightweight physical simulation engine—using real-time data such as fire source location, environmental wind field parameters (wind speed, wind direction), ambient temperature, and confined space height as input boundary conditions, and employing a simplified proxy model (such as a regional model or empirical model) of a fire dynamics simulator (FDS) to rapidly extrapolate the evolution trend of the fire within a future time window in virtual space.

[0046] Building upon this foundation, the system introduces a Physical Information Neural Network (PINN) as the core model for fire prediction. PINN embeds fire dynamics equations—including those describing hot smoke flow (Navier-Stokes equations), thermal radiation propagation (Radiative Transfer Equation), and gas diffusion (Mass Conservation Equation)—as physical constraints into the neural network's loss function. Specifically, PINN's loss function consists of two parts: a data loss term (fitting real-time thermal imaging data and gas concentration data collected by onboard sensors) and a physical loss term (penalizing the degree to which the network output deviates from the fire dynamics equations). In this way, PINN is driven by both real sensor data and constrained by physical laws during training, resulting in higher prediction accuracy and interpretability compared to purely data-driven black-box models, while still ensuring that the output conforms to physical laws even with sparse training data.

[0047] During the prediction phase, the system uses the current fire source location, heat release rate, and environmental wind field parameters as initial conditions to drive PINN to predict the fire distribution in multiple future time slices (e.g., 30 seconds, 60 seconds, 90 seconds, and 120 seconds) within the digital twin environment. It outputs the predicted radiative heat flux distribution and predicted smoke concentration distribution for each time slice, and generates a predicted thermal grid map for each time slice accordingly. When searching for evacuation paths, the path planning module no longer evaluates path costs solely based on the current thermal grid map. Instead, for each candidate path, it calculates the cumulative heat exposure risk across all time slices within the prediction time window. The formula for calculating the cumulative heat exposure risk is: R_total = Σ_t Σ_n γ^t · H( node_n , t ) Where t is the time slice index, n is the node index on the path, γ is the time decay factor (0 < γ < 1, used to balance near-term and long-term risks), and H(node_n, t) is the thermal hazard cost of node n on time slice t. The system selects the path that minimizes the cumulative thermal exposure risk over the entire process as the final avoidance path, rather than the path that minimizes the risk at the current moment.

[0048] Meanwhile, the system employs a Receding Horizon Optimization (RHorizon) strategy—updating the boundary conditions and initial state of the digital twin model every 5 seconds based on the latest sensor data, re-executing PINN prediction and path search, forming a closed loop of "prediction → planning → execution → re-prediction". Even if the fire situation changes abruptly during vehicle movement (e.g., a sudden change in wind direction or a sudden increase in the heat release rate of the fire source), RHorizon Optimization can promptly capture the changes and adjust the path in the next prediction cycle, ensuring that the vehicle always travels along the "safest trajectory for the foreseeable future". This solution upgrades path planning from reactive to proactive, effectively avoiding the problem of vehicles traveling along a currently safe path but entering a future high-temperature zone midway through the journey, making it particularly suitable for scenarios with rapidly developing fires and drastic environmental changes.

[0049] Furthermore, the specific implementation method for determining the path search cost and generating avoidance paths based on the thermal hazard level information of each grid cell in the dynamic thermal grid diagram is as follows.

[0050] In one implementation, the path planner described above can employ an improved A* algorithm for path searching. During the search process, the algorithm evaluates the actual distance g(n) from the starting point to the current node, and the estimated distance h(n) from the current node to the target node, while also incorporating thermal hazard cost h_thermal(n) and dynamic threat cost h_dynamic(n).

[0051] The thermal hazard cost is determined based on the thermal hazard level of the current node's grid: level 3 grids are impassable, level 2 grids have a maximum thermal hazard cost (e.g., 1000), level 1 grids have a medium cost (e.g., 50), and level 0 grids have a cost of 0. The dynamic threat cost is calculated based on the probability of fire spread when a vehicle arrives at the node, predicted by the fire spread rate. Higher fire source heat release rates and greater wind speeds result in higher dynamic threat costs. The expression for the comprehensive cost function is: f(n) = g(n) + α·h_thermal(n) + β·h_dynamic(n) Where α is the weighting coefficient for thermal hazard cost, and β is the weighting coefficient for dynamic threat cost. The system searches for the path with the minimum total cost, ensuring that all nodes are passable, by minimizing the total cost; this path is the generated risk avoidance path.

[0052] Furthermore, the specific implementation methods for determining the thermal hazard cost weight coefficient α and the dynamic threat cost weight coefficient β based on the current fire situation are as follows.

[0053] The current fire situation is determined by comprehensively analyzing the flame area and gas concentration data from thermal imaging. In one example, when smoke is detected but no open flame is detected, it is classified as a low-risk mode. In this case, the thermal hazard cost weighting coefficient α is set to 0.3, and the dynamic threat cost weighting coefficient β is set to 0.1, prioritizing the shortest travel distance. When a local open flame is detected and the ambient wind speed is less than a preset wind speed threshold (e.g., 5 m / s), it is classified as a medium-risk mode. In this case, α is set to 1.0, and β is set to 0.5, with the thermal hazard cost and distance cost weights being roughly equal. When a large area of ​​open flame is detected or the ambient wind speed is greater than or equal to a preset wind speed threshold (e.g., 5 m / s), it is classified as a high-risk mode. In this case, α is set to 5.0, and β is set to 2.0, with the thermal hazard cost weighting significantly higher than the distance cost, prioritizing a thermally safe path over the shortest path.

[0054] Furthermore, after generating the evacuation path in step S130, the system also monitors changes in environmental wind direction and heat release rate of the fire source in real time, and performs dynamic replanning when the triggering conditions are met.

[0055] Specifically, the system continuously monitors data from environmental wind direction sensors at a certain frequency (e.g., 50Hz), and simultaneously estimates real-time changes in the heat release rate of the fire source based on changes in the flame area in the thermal imaging image. When a wind direction change angle exceeding a preset threshold (e.g., 30°) is detected, it indicates that the fire spread direction may have changed significantly, and the existing thermal grid map cannot accurately reflect the new radiation distribution, requiring a replanning of the path. When an increase in the heat release rate of the fire source exceeding a preset percentage threshold (e.g., 20%) is detected, it indicates that the fire is intensifying and the high-temperature area is expanding rapidly, also requiring a path update.

[0056] When any of the above triggering conditions are met, the system uses the D* Lite algorithm to locally update the current path within a preset time (e.g., 100ms). The D* Lite algorithm can utilize the calculation results of the previous search and only recalculate the affected areas, thus its computational efficiency is higher than that of global replanning starting from scratch. The replanned new path replaces the original path and is sent to the execution layer to ensure that the vehicle always travels on the optimal path under the current thermal safety conditions.

[0057] It is understood that the specific values ​​mentioned above are for illustrative purposes only and are not intended to be specific. In practical applications, they can be adjusted according to the specific circumstances.

[0058] In an alternative implementation, to further reduce the computational latency of dynamic replanning and enable vehicles to quickly adapt and generate optimal routes when faced with new fire scenarios never seen before, this application also provides a fast scenario-adaptive path replanning scheme based on meta-learning. In this scheme, the system employs a Model-Agnostic Meta-Learning (MAML) algorithm framework, enabling the path planning model to pre-learn meta-knowledge of "how to quickly adapt to new scenarios" across a large number of fire scenarios, rather than simply learning the optimal strategy for a specific scenario.

[0059] Specifically, during the offline training phase, the system generates thousands of different simulated fire scenarios in a digital twin environment. Each scenario has different parameters such as fire source location, fire heat release rate, wind speed and direction, and confined space geometry. The MAML algorithm performs dual-loop training on these scenarios: the inner loop performs rapid adaptation on each scenario using a small number of gradient updates (e.g., 5 to 10 steps), while the outer loop evaluates the overall performance of the adapted model across all scenarios and updates the initial parameters of the base model. Through this dual-loop training, the base model eventually converges to a set of initial parameters that are "sensitive" to various fire scenarios—when faced with a new scenario, only a small number of gradient updates are needed to achieve near-optimal performance in that scenario.

[0060] During the real-vehicle inference phase, when a vehicle encounters a fire, the system uses current sensor data (including thermal imaging data, gas concentration data, environmental wind field parameters, vehicle position, etc.) as training samples to perform a small number (5 to 10 steps) of gradient fine-tuning on the pre-trained base model, enabling the model to quickly adapt to the unique characteristics of the current fire distribution. After adaptation, the system immediately uses the fine-tuned model to generate the current evacuation path. When changes in environmental wind direction or an increase in the heat release rate of the fire source triggers dynamic replanning conditions, the system does not need to re-execute the global path search from scratch. Instead, based on the previously fine-tuned model, it uses the latest sensor data for incremental fine-tuning (requiring only 3 to 5 gradient updates) to quickly generate the updated path.

[0061] In one specific implementation, after generating the avoidance path in step S130, different execution strategies are selected based on the occupant status inside the vehicle.

[0062] The system detects the presence of occupants inside the vehicle using seat pressure sensors and an in-vehicle camera. If no occupants are detected (e.g., the vehicle is parked and the seat pressure sensors do not detect effective pressure), the system automatically executes a hazard avoidance maneuver. It sends a hazard avoidance path to the powertrain and chassis controllers (VCU, EPS, ESP), controlling the vehicle to automatically drive to a safe area along the path. The automatic driving speed does not exceed a preset speed threshold (e.g., 10 km / h). Upon reaching the safe area, the system automatically engages Park (P) and applies the electronic parking brake. If occupants are detected, the system projects a dynamic heat map and hazard avoidance path onto the windshield using an augmented reality head-up display (AR-HUD). Simultaneously, it provides auxiliary guidance via the central control screen and voice prompts, helping the driver understand the current heat field distribution and recommended driving direction. The driver then follows the guidance to complete the hazard avoidance maneuver.

[0063] Figure 2 The path search algorithm logic block diagram for the fusion thermal cost function provided in the embodiments of this application shows the full-process control logic from system power-on to risk avoidance completion.

[0064] like Figure 2 As shown, after power-on, the system first enters a low-power monitoring state, in which the sensors are in standby mode to reduce energy consumption. When the multimodal fire sensing module detects a fire signal, the system is awakened and performs fire source localization—fusing thermal imaging data and gas concentration data to determine the location of the fire source. Subsequently, the system acquires environmental parameters (wind speed, wind direction, temperature, humidity) as input conditions for thermal radiation field modeling.

[0065] The system performs dynamic thermal radiation field modeling based on the fire source location and environmental parameters, employing vehicle-mounted CFD for simplified calculations (including a Gaussian smoke diffusion model and a point source thermal radiation model) to calculate the radiative heat flux and smoke concentration values ​​at each spatial point within the environment. The calculation results are then mapped to a gridded space, constructing an N×M grid of thermal grid maps, with each grid assigned a thermal hazard level (level 0 to 3). The path planning module reads the thermal grid map and the geometric obstacle grid map established by millimeter-wave radar, defining a comprehensive cost function f(n) = g(n) + α·h_thermal(n) + β·h_dynamic(n), where g(n) is the distance cost, h_thermal(n) is the thermal hazard cost, h_dynamic(n) is the dynamic threat cost, and α and β are weighting coefficients dynamically adjusted according to the urgency of the fire.

[0066] The A* algorithm searches for the optimal path based on the aforementioned cost function. During the search, it automatically avoids deadly grids with a thermal hazard level of 3 and high-cost grids with a thermal hazard level of 2, selecting the path with the lowest total cost as the avoidance path. During path execution, the system continuously monitors changes in wind direction and the heat release rate of the fire source. Dynamic replanning is triggered when a change in wind direction exceeds a preset angle or an increase in the heat release rate of the fire source exceeds a preset proportion. Finally, after the avoidance is completed, the status is reported to the vehicle owner's app or cloud platform, and the process ends.

[0067] Figure 3 The flowchart of the path planning based on the fusion thermal cost function provided in this application embodiment shows the core logic of the path search module in generating the optimal safe path based on the thermal grid map, geometric obstacle map and origin and destination information.

[0068] like Figure 3 As shown, the input to the path planning module includes three parts: a dynamic heat map, a geometric obstacle map, and start and end point location information. The dynamic heat map contains thermal hazard level information for each grid cell, the geometric obstacle map contains physical obstacle information detected by millimeter-wave radar, the start point is the vehicle's current position, and the end point is the pre-selected safe zone location.

[0069] The path planning module employs an improved A* algorithm for path search. During the search process, the algorithm dynamically adjusts the weighting coefficients α and β in the comprehensive cost function f(n) = g(n) + α·h_thermal(n) + β·h_dynamic(n) based on the current fire situation. When the fire is large or the wind speed is high, the system dynamically increases the values ​​of α and β, making the path planning more biased towards thermal safety rather than the shortest distance. For each path point expanded during the search process, the algorithm performs a feasibility judgment—checking whether the thermal radiation flux of the grid where the path point is located is lower than the thermal failure threshold of vehicle components (tires, high-voltage wiring harnesses, etc.) and whether the smoke concentration is lower than the lethal threshold. If the radiative heat flux or smoke concentration of the path point exceeds the corresponding threshold, the path point is marked as impassable, and the algorithm will not include it in the candidate paths.

[0070] After comprehensively evaluating each candidate path in the search space, the algorithm outputs the path with the minimum total cost as the optimal safe path. This path geometrically avoids physical obstacles and thermodynamically avoids high-temperature and flue gas areas, achieving optimal path planning under dual safety constraints.

[0071] The technical solution of this application will be fully explained below using a specific scenario of a neighboring car catching fire in an underground parking garage.

[0072] In a parking garage on level B2, a gasoline-powered car spontaneously combusted due to aging electrical wiring in a row of parallel parking spaces. An adjacent pure electric vehicle equipped with this system was parked unoccupied. After the neighboring car caught fire, it produced a large amount of smoke. Aerosol sensors located around the electric vehicle detected a rapid increase in CO concentration from the background value of 10 ppm to 85 ppm, triggering the system's activation.

[0073] The system activates an infrared thermal imaging camera for secondary confirmation. The camera captures a high-temperature area in the direction of the adjacent vehicle. A lightweight CNN model segments the thermal image to identify the core area of ​​the flame. Simultaneously, the concentration differences from multiple aerosol sensors form a concentration gradient direction. However, the ventilation airflow in the underground parking garage causes this gradient to drift. The system uses a particle filter algorithm to fuse the location of the high-temperature area identified by the thermal imaging with the gas concentration gradient direction, correcting the location drift caused by airflow interference and accurately determining the three-dimensional coordinates of the fire source center in the vehicle's coordinate system.

[0074] The system acquires environmental parameters for the underground parking garage: wind speed approximately 0.5 m / s, wind direction variable, and air temperature 22℃. Since the underground parking garage is a confined space, the system uses a floor height of 3.5 m and calculates the Froude number Fr≈0.085, correcting the smoke diffusion coefficient in the physical diffusion model. The system divides the environment around the vehicle into grids (20cm×20cm resolution). Based on the fire source location and the corrected physical diffusion model, it calculates the radiative heat flux and smoke concentration at the center of each grid, comparing these values ​​with preset vehicle component thermal failure thresholds and smoke concentration lethal thresholds to generate a dynamic thermal grid map. Grids near the fire source in the underground parking garage are marked as Level 3 (lethal), progressing outwards to Level 2 (dangerous), Level 1 (warning), and Level 0 (safe).

[0075] The planner reads the dynamic thermal grid map and the geometric obstacle grid map established by millimeter-wave radar, and performs an improved A* path search. Due to the cost of thermal hazards, the algorithm abandons the geometrically shortest path that crosses a Level 2 danger zone, and chooses a path to detour to a Level 0 safety zone. The system detects no occupants in the vehicle and automatically performs an avoidance maneuver. The vehicle autonomously drives away along the planned path at a speed of 8 km / h, targeting a safe area more than 25 meters from the fire source center. During the vehicle's movement, the thermal imaging camera continuously updates the fire source location and the thermal grid map. When a wind direction change exceeding 30° is detected, the system triggers local replanning, adjusting the remaining path to avoid newly formed high-temperature areas. After reaching the safe area, the vehicle automatically engages Park (P) gear, applies the electronic parking brake, and sends an avoidance completion notification and the vehicle's new location information to the owner's app via T-Box.

[0076] Through the above process, this application actively wakes up the system after a fire occurs, and through multimodal perception cross-positioning, dynamic thermal field modeling and thermal safety path search, enables the vehicle to automatically drive away from the high-temperature area without passengers, effectively preventing the vehicle from being affected by the fire.

[0077] This application determines the fire source location by acquiring thermal imaging data and gas concentration data and cross-validating them. It utilizes thermal imaging to identify high-temperature areas and gas concentration gradients to correct for location drift caused by airflow interference, improving the accuracy of fire source location and avoiding the problem of inaccurate positioning by a single sensor under smoke or airflow interference. Based on the fire source location and environmental wind field parameters, it calculates radiative heat flux using a physical diffusion model and generates a dynamic thermal grid map. This allows the path planning input to no longer be limited to static geometric obstacles but to include dynamic physical field information on thermal radiation and smoke diffusion, solving the problem that static obstacle avoidance cannot cope with fire spread. By incorporating thermal hazard level information as part of the path search cost to generate an avoidance path, the application ensures that path planning considers thermal safety while performing geometric obstacle avoidance, ultimately outputting a driving path that avoids both physical obstacles and high-temperature smoke. Overall, this application represents an upgrade from static geometric obstacle avoidance to dynamic thermodynamic safety path planning, effectively preventing secondary injuries caused by vehicles entering high-temperature smoke areas during hazard avoidance.

[0078] Based on the above method embodiments, this application also provides a vehicle fire escape path generation device, see [link to relevant documentation]. Figure 4 As shown, the device includes the following parts: The fire source location module 410 is used to acquire thermal imaging data and gas concentration data of the environment around the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data. The thermal field modeling module 420 is used to calculate the radiative heat flux value of each spatial point in the environment space according to the fire source location and environmental wind field parameters through a preset physical diffusion model, and to map the radiative heat flux value to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information. The path planning module 430 is used to determine the path search cost based on the thermal hazard level information of each grid unit in the dynamic thermal grid map, and to generate a risk avoidance path to a safe area through the path search cost.

[0079] The device provided in this application embodiment has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts of the device embodiment not mentioned can be referred to the corresponding content in the aforementioned method embodiment.

[0080] This application also provides a vehicle fire escape path generation system, which mainly includes: a hazard perception module for collecting thermal imaging data and gas concentration data of the environment around the vehicle; a central processing and decision-making unit connected to the hazard perception module for receiving the thermal imaging data and gas concentration data and executing the vehicle fire escape path generation method as described in any of the foregoing embodiments to generate an escape path to a safe area; and a vehicle execution control module connected to the central processing and decision-making unit for receiving the escape path and controlling the vehicle to travel according to the escape path.

[0081] Specifically, Figure 5 This paper illustrates the specific structure of a vehicle fire escape path generation system, demonstrating efficient collaboration among multiple modules at both the hardware and algorithm levels. The front end is a hazard perception module, which does not rely on a single sensor but integrates an infrared thermal imaging device, millimeter-wave radar, aerosol sensors, and environmental sensors. The infrared thermal imaging captures the distribution of thermal radiation and uses the U-Net algorithm for image segmentation and accurate temperature calculation. The millimeter-wave radar detects the spatial occupancy of surrounding obstacles, while the aerosol and environmental sensors monitor CO / CO2 concentrations, wind speed, wind direction, temperature, and other external conditions in real time. This provides a multi-source and complementary physical perception foundation for subsequent decision-making.

[0082] In the central processing and decision-making unit, the system first uses a particle filtering algorithm to deeply fuse the aforementioned multi-source sensing data, thereby accurately determining the true coordinates of the fire source in three-dimensional space. After obtaining the fire source location, the system does not stop at static hazard assessment, but instead constructs a dynamic thermal radiation field in real time based on the Stefan-Boltzmann thermal radiation law, combined with a high-temperature smoke diffusion model and Froude number correction. This thermal field is further transformed into gridded data, and the thermal hazard level is divided into 0 to 3, and the specific radiative heat flux is calculated. In the path generation stage, the system adopts an improved A* path search algorithm, which extends the traditional distance cost function to incorporate a dynamic cost function that incorporates the physical threat of thermal radiation, ensuring that the planned avoidance path is both geometrically feasible and avoids high-temperature and smoke areas to the greatest extent possible.

[0083] The system is equipped with comprehensive human-machine interaction and remote communication interfaces. On one hand, through a HUD or large screen, the system can render dynamic thermal fields and display path previews to drivers and passengers, maintaining information transparency. On the other hand, the vehicle-mounted T-Box communication module reports hazard data, fire coordinates, and control requests to the remote terminal in real time. At the cloud level, the remote terminal integrates vehicle-cloud collaboration capabilities, enabling it to leverage edge cloud CFD computing resources to perform secondary calculations of smoke diffusion and, combined with real-time weather data from the meteorological API interface, provide cloud-based optimization and correction of the path planned by the vehicle.

[0084] At the system's core, the central processing unit (CPU) translates the finalized hazard avoidance path into specific control requests and sends them to the vehicle's execution control module. This module typically includes EPB (Electronic Parking Brake), ESP (Electronic Stability Program), EPS (Electric Power Steering), and a powertrain control system. As the vehicle travels along the predetermined hazard avoidance path, the execution module continuously feeds back the vehicle's actual operating status to the central decision-making unit, forming a closed loop of "perception-decision-control-feedback." If a sudden hazard or change in environmental variables is detected during travel, the CPU can quickly reactivate the path search and adjust the hazard avoidance route in real time, thereby achieving safe and intelligent emergency control of the vehicle in a fire scenario.

[0085] This application also provides an electronic device, such as... Figure 6 The diagram shows the structure of the electronic device 100, which includes a processor 61 and a memory 60. The memory 60 stores computer-executable instructions that can be executed by the processor 61. The processor 61 executes the computer-executable instructions to implement any of the methods described above.

[0086] exist Figure 6 In the illustrated embodiment, the electronic device further includes a bus 62 and a communication interface 63, wherein the processor 61, the communication interface 63, and the memory 60 are connected via the bus 62.

[0087] The memory 60 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 63 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 62 may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 62 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0088] Processor 61 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 61 or by instructions in software form. Processor 61 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in the memory, and the processor 61 reads the information in the memory and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiment.

[0089] This application also provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions cause the processor to implement the above-described method. For specific implementation details, please refer to the foregoing method embodiments, which will not be repeated here.

[0090] The computer program product of the vehicle fire escape path generation method, device and electronic device provided in the embodiments of this application includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.

[0091] Unless otherwise specifically stated, the relative steps, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0092] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0093] In the description of this application, it should be noted that the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0094] In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "set up," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for generating vehicle fire escape routes, characterized in that, include: Acquire thermal imaging data and gas concentration data of the environment surrounding the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data; Based on the fire source location and environmental wind field parameters, the radiative heat flux value of each spatial point in the environmental space is calculated through a preset physical diffusion model, and the radiative heat flux value is mapped to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information. Based on the thermal hazard level information of each grid cell in the dynamic thermal grid, the path search cost is determined, and an escape path to a safe area is generated using the path search cost.

2. The method according to claim 1, characterized in that, Determining the location of the fire source based on the thermal imaging data and gas concentration data includes: The thermal imaging data is processed for image segmentation to identify high-temperature areas; Calculate the concentration gradient direction of the gas concentration data; The location of the high-temperature region is corrected based on the direction of the concentration gradient to determine the location of the fire source.

3. The method according to claim 1, characterized in that, Based on the fire source location and environmental wind field parameters, the radiative heat flux value of each spatial point in the environmental space is calculated using a preset physical diffusion model. This radiative heat flux value is then mapped to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information, including: Based on the location of the fire source and the environmental wind field parameters, the radiative heat flux and smoke concentration values ​​at each spatial point in the environment are calculated using a preset physical diffusion model. The radiant heat flux value is compared with a preset vehicle component thermal failure threshold, and the flue gas concentration value is compared with a preset environmental safety threshold. Based on the comparison results, the thermal hazard level of each spatial point is determined, and the spatial points with the thermal hazard level are mapped to the rasterized space to generate the dynamic thermal raster map.

4. The method according to claim 3, characterized in that, Based on the location of the fire source and the environmental wind field parameters, the radiative heat flux and smoke concentration at each point in the environmental space are calculated using a preset physical diffusion model, including: Obtain the height parameters of the confined space in the vehicle's environment; The Froude number is calculated based on the environmental wind field parameters and the confined space height parameters. The preset physical diffusion model is modified based on the fire source location and the Froude number to calculate the radiative heat flux and smoke concentration at each spatial point in the environment.

5. The method according to claim 1, characterized in that, Based on the thermal hazard level information of each grid cell in the dynamic thermal grid map, the path search cost is determined, and an escape path to a safe area is generated using the path search cost, including: The distance cost is determined based on the geometric distance between nodes during the path search process, the thermal hazard cost is determined based on the thermal hazard level information in each grid cell of the dynamic thermal grid map, and the dynamic threat cost is determined based on the urgency of the fire. The first weighting coefficient corresponding to the thermal hazard cost and the second weighting coefficient corresponding to the dynamic threat cost are determined based on the current fire situation. The distance cost, the thermal hazard cost under the first weighting coefficient, and the dynamic threat cost under the second weighting coefficient are input into a preset comprehensive cost function for calculation. By minimizing the total value, a risk avoidance path to the safe zone is generated.

6. The method according to claim 5, characterized in that, Determine the first weighting coefficient corresponding to the thermal hazard cost and the second weighting coefficient corresponding to the dynamic threat cost based on the current fire situation, including: When a local open flame is detected and the ambient wind speed is less than a preset wind speed threshold, it is determined to be a medium-risk mode, and the first weight coefficient and the second weight coefficient corresponding to the medium-risk mode are determined respectively. When a large-area open flame is detected or the ambient wind speed is greater than or equal to the preset wind speed threshold, it is determined to be a high-risk mode, and the first weight coefficient and the second weight coefficient corresponding to the high-risk mode are determined respectively. In the high-risk mode, both the first weighting coefficient and the second weighting coefficient are greater than those in the medium-risk mode.

7. The method according to claim 1, characterized in that, After generating an evacuation path to a safe zone using the path search cost, the method further includes: Real-time monitoring of changes in ambient wind direction and heat release rate from fire sources; When the detected wind direction change angle exceeds a preset angle threshold or the heat release rate of the fire source increases beyond a preset proportion threshold, the evacuation path is dynamically replanned.

8. The method according to claim 1, characterized in that, After generating an evacuation path to a safe zone using the path search cost, the method further includes: Detect the occupant status inside the vehicle; If no occupants are detected inside the vehicle, the vehicle will be controlled to automatically travel to the safe area according to the described avoidance path. If an occupant is detected inside the vehicle, the dynamic thermal grid map and the hazard avoidance path are projected onto the windshield via an augmented reality head-up display device, assisting the driver in following the hazard avoidance path.

9. A vehicle fire escape route generation device, characterized in that, include: The fire source location module is used to acquire thermal imaging data and gas concentration data of the environment around the vehicle, and determine the location of the fire source based on the thermal imaging data and gas concentration data. The thermal field modeling module is used to calculate the radiative heat flux value of each spatial point in the environment space according to the fire source location and environmental wind field parameters through a preset physical diffusion model, and to map the radiative heat flux value to a gridded space to generate a dynamic thermal grid map containing thermal hazard level information. The path planning module is used to determine the path search cost based on the thermal hazard level information of each grid cell in the dynamic thermal grid map, and to generate an avoidance path to a safe area based on the path search cost.

10. A vehicle fire escape route generation system, characterized in that, include: The hazard perception module is used to collect thermal imaging data and gas concentration data of the vehicle's surrounding environment; The central processing and decision-making unit is connected to the hazard perception module and is used to receive the thermal imaging data and gas concentration data, and execute the vehicle fire evacuation path generation method as described in any one of claims 1 to 8 to generate an evacuation path to a safe area. The vehicle execution control module, connected to the central processing and decision-making unit, is used to receive the avoidance path and control the vehicle to travel according to the avoidance path.

11. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 8.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method according to any one of claims 1 to 8.

Images