Intelligent alarm method for driving intervention after disaster

Abstract

The invention discloses an intelligent alarm method for driving intervention after a disaster, and the method comprises the steps: collecting a visible light image and an infrared thermal imaging image, and scanning the coordinate data of a three-dimensional space at the same time; inputting the visible light image into a convolutional neural network model to perform road obstacle type identification, and generating three-dimensional contour data of the obstacle in combination with the three-dimensional space coordinate data; when the height of the obstacle exceeds a height threshold value and the distance between the obstacle and the vehicle is smaller than a distance threshold value, uploading a feature data set containing the GPS coordinates, the three-dimensional contour data and the obstacle type to a cloud server through a cellular network; the cloud server matches the feature data set with a pre-stored disaster area digital elevation model based on the three-dimensional space coordinate data, and calculates the ratio of the projection area of the obstacle on the current lane cross section to the lane width; and when the ratio exceeds a ratio threshold value, generating an alarm instruction. According to the invention, the accident risk caused by misjudgment of obstacles or improper path planning after disasters can be effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things. More specifically, the present invention relates to an intelligent alarm method for driving intervention after a disaster. Background Art

[0002] After a disaster occurs, the road environment often becomes complex and dangerous, posing many challenges to driving safety. Traditional driving safety alarm systems have certain limitations when dealing with post-disaster scenarios. Existing vehicle-mounted obstacle detection systems mainly rely on a single sensor, such as a camera or a millimeter-wave radar. However, a single sensor has obvious deficiencies in a disaster environment. For example, a camera is greatly affected by light, smoke, dust, etc., and its detection ability significantly decreases at night or in bad weather conditions; although a millimeter-wave radar can provide three-dimensional space coordinate data, its ability to identify obstacle types is limited and it cannot accurately distinguish obstacles of different materials and characteristics. This limitation of a single sensor results in insufficient reliability and accuracy of the system in the complex post-disaster environment. Moreover, after a disaster occurs, the geological conditions of the road may change, such as landslides, ground subsidence, etc. The traditional lane model is fixed and cannot adapt to this dynamically changing geological environment. When a vehicle is driving on a section with geological hazards, the fixed lane model cannot provide an accurate risk assessment, which may lead to inappropriate avoidance path selection.

[0003] Therefore, it is necessary to design a technical solution that can overcome the above defects. Summary of the Invention

[0004] An object of the present invention is to provide an intelligent alarm method for driving intervention after a disaster, which can effectively reduce the accident risk caused by misjudgment of obstacles or improper path planning after a disaster.

[0005] To achieve these objects and other advantages of the present invention, according to one aspect of the present invention, there is provided an intelligent alarm method for driving intervention after a disaster, including: S1: Collect visible light images and infrared thermal imaging images within a range of 0 - 150 meters in front of the vehicle through an in-vehicle multi-spectral camera, and at the same time scan three-dimensional spatial coordinate data within a range of 0 - 150 meters in front through an in-vehicle millimeter-wave radar; S2: Input the visible light image into a convolutional neural network model for road obstacle type recognition, and at the same time perform temperature gradient analysis on the infrared thermal imaging image, and generate three-dimensional contour data of the obstacle in combination with the three-dimensional spatial coordinate data of the millimeter-wave radar; S3: When the height of the obstacle exceeds the height threshold and the distance from the vehicle is less than the distance threshold, call the vehicle positioning module to obtain the current GPS coordinates, and upload a feature data set including the GPS coordinates, the three-dimensional contour data, and the obstacle type to the cloud server through the cellular network; S4: Based on the three-dimensional spatial coordinate data, the cloud server performs spatial matching between the feature data set and a pre-stored digital elevation model of the disaster area, and calculates the ratio of the projected area of the obstacle on the cross-section of the current lane to the lane width; S5: When the ratio exceeds the ratio threshold, generate an alarm instruction including the three-dimensional coordinates of the obstacle, the obstacle type, and avoidance path parameters, and send the alarm instruction to the vehicle through the cellular network.

[0006] Further, S2 further includes: S21: Based on the temperature values of each pixel point in the infrared thermal imaging image, divide multiple temperature intervals and mark them as different color channels; S22: Identify continuous temperature anomaly regions. If the temperature difference between a certain region and its surroundings exceeds the dynamic threshold and lasts for more than 3 seconds, it is determined as residues of thermal radiation disasters or low-temperature icing areas; S23: Perform spatial overlay of the obstacle bounding box recognized by the convolutional neural network in the visible light image and the temperature anomaly region. When the overlap degree exceeds 80%, activate the multi-modal verification mechanism and reconstruct the surface deformation characteristics of the obstacle through millimeter-wave radar point cloud data; S24: According to the correspondence between the surface deformation characteristics and the temperature distribution, correct the geometric distortion caused by optical occlusion in the three-dimensional contour data.

[0007] Further, locate the spatial feature point set of the obstacle through millimeter-wave radar point cloud data. If the distance between the temperature anomaly region in the infrared thermal imaging and the centroid of the feature point set is less than 0.5 meters, mark it as a credible thermal radiation region; within the credible thermal radiation region, identify the boundary where the texture of the visible light image is continuous but the radar point cloud density drops suddenly as the optical occlusion boundary; according to the relationship between the temperature gradient direction and the change in radar point cloud density, select an interpolation algorithm to reconstruct the three-dimensional contour within the occlusion boundary: when the high temperature extends towards the sparse direction of the point cloud, use the depression repair algorithm to fill the contour; when the high temperature extends towards the dense direction of the point cloud, use the protrusion repair algorithm to correct the contour.

[0008] Further, S3 further includes: S31: Before invoking the vehicle positioning module, judge the material properties of obstacles according to the reflection intensity data of the millimeter-wave radar, and set the priority of metal obstacles to the highest; S32: When the cellular network bandwidth is lower than the preset value, start the data compression engine: adopt differential coding compression for the three-dimensional contour data of consecutive frames, and retain the original data of the highest temperature area and compress the background area for the infrared image; S33: Dynamically select the target server cluster for uploading according to the disaster type database, and send landslide data to the geological disaster monitoring server and debris flow data to the hydrological monitoring server.

[0009] Further, S4 specifically includes: S41: Extract the historical deformation parameters of the road ahead of the vehicle from the digital elevation model of the disaster area, and calculate the geological settlement trend coefficient of the projection position of the current obstacle; S42: Establish a virtual lane expansion model, superimpose the influence amount of the geological settlement trend coefficient on the lane width reference value, and generate a dynamic lane width reference value; S43: Perform stress deformation simulation on the three-dimensional contour of the obstacle along the driving direction, and predict the maximum expansion value of the projected area within the next 10 seconds; S44: Use the real-time ratio of the dynamic lane width reference value to the maximum expansion value as the spatial matching index.

[0010] Further, S41 specifically includes: S411: Extract the surface elevation change data of the road ahead of the vehicle per hour within the most recent 72 hours from the digital elevation model of the disaster area, and construct a time-series deformation map; S412: Identify periodic uplift or settlement areas in the deformation map. If the deformation rate of the same coordinate point exceeds the threshold for 3 consecutive hours and the change direction is the same, mark it as an active deformation point; S413: Establish a dynamic monitoring grid with a radius of 5 meters centered on the projection position of the obstacle, and count the proportion of active deformation points and the average deformation acceleration within the grid; S414: Dynamically calculate the geological settlement trend coefficient according to the weighted value of the proportion of active deformation points and the average deformation acceleration, where: when the proportion of active deformation points exceeds 40% and the acceleration is positive, the coefficient value is adjusted up to 1.2 - 1.5 times the reference value; when the proportion of active deformation points is lower than 10% and the acceleration is negative, the coefficient value is adjusted down to 0.6 - 0.8 times the reference value.

[0011] Further, the S43 specifically includes: S431: Identify the surface texture features of the obstacle through the vehicle-mounted multi-spectral camera, and classify it as a rigid material or a flexible material in combination with the corrected three-dimensional contour data; S432: Apply a virtual wind resistance load in the driving direction, and the value of the wind resistance load is dynamically calculated according to the ratio of the current vehicle speed to the obstacle height; S433: If it is a rigid material, monitor the infrared temperature change rate at the edge of the obstacle base, and activate the thermal expansion compensation model when the rate exceeds 2 °C / second, and add an axial extension margin to the projected area calculation; S434: If it is a flexible material, retrieve the current precipitation data and simulate the rainwater penetration and softening effect, and calculate the envelope range of the deformation and expansion within the next 10 seconds based on the material porosity parameter; S435: Take the maximum superposition value of the thermal expansion compensation margin and the softening effect envelope range as the maximum expansion value.

[0012] Further, the S5 includes: S51: Retrieve the avoidance trajectory data of all vehicles on the current section in the past 24 hours from the cloud server to construct a probability density distribution heat map; S52: According to the vehicle wheelbase parameter and the lateral extension length of the obstacle, screen the candidate paths that meet the vehicle kinematic constraints in the heat map; S53: Prioritize the path with the highest coincidence degree with the geological stable zone in the digital elevation model of the disaster area among the candidate paths; S54: Decompose the selected path into a longitudinal deceleration curve and a lateral offset instruction sequence, where the lateral offset instruction includes a steering angle safety margin value calculated in real time based on the tire grip.

[0013] Further, the S53 includes: S531: Extract the strike data of the rock fracture surface from the digital elevation model of the disaster area, and mark the fracture zone with an angle less than 30 degrees with the driving direction as a potentially unstable area; S532: Obtain the ground vibration monitoring data of the current section in the past 2 hours. If a certain candidate path passes through a grid cell with a vibration amplitude exceeding 0.3g and a duration exceeding 10 seconds, the geological stability coefficient of this path will be downgraded; S533: Calculate the surface bearing capacity margin of each candidate path according to the vehicle full load mass parameter, and eliminate the paths with a margin value lower than the safety threshold; S534: Weight and sort the remaining paths according to the fracture zone avoidance rate and the surface bearing capacity margin, and select the path with the highest comprehensive score.

[0014] Further, the S54 includes: S541: obtaining the real-time monitored value of the tire pressure, the infrared temperature value of the road surface, and the road surface roughness index detected by the millimeter-wave radar; S542: when the tire pressure is 15% lower than the standard value and the road surface temperature is higher than 50 °C, activating the high-temperature soft tire compensation mode and increasing the basic safety margin by 20%; S543: if the road surface roughness index continuously drops by more than 40% for 5 consecutive frames, determining it as a slippery road surface and injecting a sine damping factor into the lateral offset command to suppress steering jitter; S544: dynamically adjusting the safety margin according to the real-time yaw angular velocity feedback of the vehicle: when the yaw angular velocity exceeds 0.5 rad / s, an additional 5% margin is added for every increase of 0.1 rad / s; when the yaw angular velocity is lower than 0.2 rad / s for 3 seconds, restoring the reference margin value.

[0015] The present invention at least includes the following beneficial effects: Through multi-sensor fusion and intelligent processing, the present invention improves the accuracy and reliability of driving safety intervention after disasters. By using an in-vehicle multi-spectral camera to collect visible light and infrared thermal imaging images, combined with the three-dimensional coordinate data of the millimeter-wave radar, multi-dimensional perception of obstacles within 150 meters ahead is achieved, solving the problem of insufficient recognition ability of a single sensor due to illumination and material, and providing comprehensive data support for subsequent analysis. By using a convolutional neural network to identify the type of obstacles, combined with infrared thermal imaging temperature gradient analysis and radar data to generate a three-dimensional contour, joint modeling of geometric features and physical properties is realized, accurately distinguishing disaster residues, avoiding geometric distortion caused by optical occlusion, and improving the accuracy of obstacle feature extraction. The cloud processing module dynamically calculates the ratio of the projected area of the obstacle to the lane width based on the digital elevation model of the disaster area, breaking through the limitation of traditional fixed threshold evaluation, adapting to complex scenarios such as road deformation and geological settlement after disasters, and ensuring that the risk judgment conforms to the actual road conditions. When the risk exceeds the threshold, the system generates an alarm command including the coordinates, type, and avoidance path parameters of the obstacle, providing clear intervention guidance for driving, effectively reducing the accident risk caused by misjudgment of obstacles or improper path planning, and constructing an efficient safety closed-loop of "perception - analysis - decision".

[0016] Other advantages, objectives, and features of the present invention will be partially reflected by the following description, and partially will also be understood by those skilled in the art through the research and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following further describes the present invention in detail with reference to the drawings, so that those skilled in the art can implement it according to the description in the specification.

[0019] It should be understood that terms such as "having", "including", and "comprising" used in the embodiments of the present application do not exclude the presence or addition of one or more other elements or combinations thereof. All directional indications (such as up, down, left, right, front, back...) in the embodiments of the present application are only used to explain the relative positional relationship and movement conditions between components in a specific posture. If the specific posture changes, the directional indications will also change accordingly. When an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or there may be an intermediate element present at the same time. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or it can also be indirectly connected to the other element through an intermediate element. The descriptions in the embodiments of the present application related to "first", "second", etc. are only for descriptive purposes and should not be construed as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature.

[0020] It should be noted that the technical solutions between the various embodiments of the present application can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present application.

[0021] As Figure 1 shown, the embodiments of the present application provide an intelligent alarm method for post-disaster driving intervention, including: collecting visible light images and infrared thermal imaging images within a range of 0 - 150 meters in front of the vehicle through an in-vehicle multi-spectral camera, and simultaneously scanning three-dimensional spatial coordinate data within a range of 0 - 150 meters in front through an in-vehicle millimeter-wave radar; inputting the visible light images into a convolutional neural network model for road obstacle type recognition, and simultaneously performing temperature gradient analysis on the infrared thermal imaging images, and generating three-dimensional contour data of the obstacle in combination with the three-dimensional spatial coordinate data of the millimeter-wave radar; when the height of the obstacle exceeds the height threshold and the distance from the vehicle is less than the distance threshold, calling the vehicle positioning module to obtain the current GPS coordinates, and uploading a feature data set including the GPS coordinates, three-dimensional contour data, and obstacle type to the cloud server through the cellular network; the cloud server, based on the three-dimensional spatial coordinate data, performs spatial matching between the feature data set and the pre-stored digital elevation model of the disaster area, and calculates the ratio of the projected area of the obstacle on the cross-section of the current lane to the lane width; when the ratio exceeds the ratio threshold, generating an alarm instruction including the three-dimensional coordinates of the obstacle, the obstacle type, and avoidance path parameters, and sending the alarm instruction to the vehicle through the cellular network.

[0022] Exemplarily, the in-vehicle multi-spectral camera can select the FLIR Boson 320 multi-spectral camera, which has a visible light and infrared dual-channel. The visible light resolution is 1920×1080, and the infrared thermal imaging resolution is 320×256. It is installed at the center above the front windshield of the vehicle, and the optical axis is aligned with the vehicle's central axis. The millimeter-wave radar can select the Bosch MRR evo2 mid-range radar, with a scanning range of 0 - 160 meters and an angular resolution of ±3.5°. It is installed at the center inside the front bumper, maintaining a height of 0.8 meters from the ground. The convolutional neural network model adopts the ResNet-50 structure, and the input image size is adjusted to 224×224. The training dataset contains 20 types of obstacle images such as falling rocks, trees, water accumulation, and ruins in disaster scenarios. After data augmentation, there are a total of 50,000 samples. The model is fine-tuned for disaster scenarios based on pre-training on ImageNet. The temperature gradient analysis of the infrared thermal imaging image generates a gradient matrix by calculating the absolute value of the temperature difference between each pixel and its 8-neighbor pixels. The temperature range is divided into 0 - 40°C (interval 10°C), 40 - 80°C (interval 20°C), and above 80°C (interval 30°C), which are respectively mapped to the blue, yellow, and red channels. When generating the three-dimensional contour data, first, the camera internal parameters are obtained through the Zhang Zhengyou calibration method, and the timestamp synchronization data of the millimeter-wave radar is used. The transformation relationship between the radar coordinate system and the camera coordinate system is established through hand-eye calibration. The radar point cloud is projected onto the image plane, and the obstacle coordinate points in the point cloud are extracted by combining the semantic segmentation results of the visible light image. The three-dimensional surface mesh is constructed through the Delaunay triangulation algorithm. The digital elevation model of the disaster area is obtained by unmanned aerial vehicle aerial survey or geological survey, stored in the DEM format, with a spatial resolution of 1 meter, containing longitude and latitude, elevation values, and road centerline vector data. The coordinate system adopts the WGS84 geographic coordinate system. The height threshold is set to 1.2 meters, the distance threshold is set to 50 meters, and the ratio threshold is set to 0.3. The cellular network module adopts the Quectel M95 4G module, which supports the TCP / IP protocol. Before data upload, it is encapsulated in JSON format, including the timestamp, GPS coordinates, three-dimensional contour vertex coordinates, and obstacle type labels. During the spatial matching process, the server extracts the road centerline within 150 meters in front of the vehicle's current position from the digital elevation model. Taking the obstacle projection position as the center, a 10-meter-long section is intercepted along the road direction to generate a cross-sectional plane perpendicular to the road centerline. For each vertex coordinate of the obstacle's three-dimensional contour, its two-dimensional projection point on the cross-section is calculated through the plane projection transformation matrix to form a closed polygon contour. The projection area is calculated using Green's formula by performing an integral operation on the polygon vertex coordinate sequence to obtain the projection area of the obstacle on the cross-section. The lane width data is derived from the road attribute table of the digital elevation model, and the reference value is the national standard lane width of 3.5 meters.

[0023] In this embodiment, the multispectral camera and the millimeter-wave radar collect data synchronously. After time alignment and spatial calibration, the visible light image is input into a convolutional neural network to identify the type of obstacle. The infrared image is used to assist in judging the material characteristics through temperature gradient analysis, and the radar point cloud data is used to construct a three-dimensional contour. When the height and distance of the obstacle meet the threshold conditions, the system packages the feature data and uploads it to the cloud. The cloud uses a digital elevation model to convert the three-dimensional coordinates into a road cross-section projection and calculates the occupancy ratio. This method realizes the efficient detection and risk assessment of obstacles after disasters through multimodal data fusion and precise spatial matching, providing a real-time and reliable basis for intervention in driving safety. In another embodiment, the processing of the infrared thermal imaging image in the above method further includes: dividing multiple temperature intervals and marking them as different color channels based on the temperature values of each pixel point in the infrared thermal imaging image; identifying continuous temperature anomaly regions, and if the temperature difference between a certain region and its surroundings exceeds the dynamic threshold and lasts for more than 3 seconds, it is determined as a residue of a thermal radiation disaster or a low-temperature icing area; spatially superimposing the obstacle bounding box recognized by the convolutional neural network in the visible light image with the temperature anomaly region, and activating the multimodal verification mechanism when the overlap degree exceeds 80%, and reconstructing the surface deformation characteristics of the obstacle through the millimeter-wave radar point cloud data; correcting the geometric distortion caused by optical occlusion in the three-dimensional contour data according to the corresponding relationship between the surface deformation characteristics and the temperature distribution.

[0024] Exemplarily, the temperature range of the infrared thermal imaging image is divided by the equal-distance segmentation method: 0 - 20°C (blue channel), 20 - 40°C (green channel), 40 - 60°C (yellow channel), above 60°C (red channel). The temperature value range corresponding to each channel is displayed in pseudo-color on the image. The dynamic threshold is calculated as the standard deviation of the temperature of the central pixel in the current area and the temperatures of the pixels in the surrounding 5×5 neighborhood. When the absolute value of the temperature difference exceeds 1.5 times the standard deviation, it is determined as abnormal. Continuous detection requires 15 consecutive frames (frame rate 5fps, corresponding to 3 seconds) to meet the conditions. The identification of the temperature anomaly area is achieved through the connected component analysis algorithm, and the minimum bounding rectangle is extracted as the area boundary. The obstacle bounding box of the visible light image is output by the convolutional neural network, in the format of (x1, y1, x2, y2). When performing spatial overlay, the infrared image and the visible light image are first registered through SIFT feature matching, and the intersection over union is calculated. When the intersection over union ≥ 80%, multi-modal verification is triggered. When reconstructing the surface deformation features from the millimeter-wave radar point cloud data, first, the noise is reduced through voxel grid filtering, then the normal vector and curvature value of each point are calculated, and the edge points with curvature mutations are extracted as the deformation feature points. Geometric distortion correction uses the weighted moving least squares method. For the missing points in the occluded area, guided by the temperature gradient direction, the coordinates of 5 adjacent valid points are weighted and averaged, and the iteration is optimized until the point cloud density of the reconstructed area reaches more than 80% of the surrounding area. The infrared camera can select FLIR T400, which is installed side by side with the visible light camera (Sony IMX490) at the front of the vehicle. The distance between the two is 10 cm, and hardware triggering is used for synchronous exposure.

[0025] In this embodiment, after the system divides the temperature range of the infrared image, it continuously detects the temperature anomaly area through the dynamic threshold, and combines the spatial overlay judgment of the visible light obstacle bounding box to activate the fine reconstruction of the surface deformation from the radar point cloud data. This process effectively identifies the thermal radiation residues or icing areas. Through the cross-validation of multi-modal data, it corrects the contour distortion caused by optical occlusion, improves the integrity and accuracy of the three-dimensional model of the obstacle, and provides more reliable geometric and physical feature information for subsequent risk assessment.

[0026] In another embodiment, the spatial feature point set of the obstacle is located through the millimeter-wave radar point cloud data. If the distance between the temperature anomaly area in the infrared thermal imaging and the centroid of the feature point set is less than 0.5 meters, it is marked as a credible thermal radiation area; within the credible thermal radiation area, the boundary where the texture of the visible light image is continuous but the radar point cloud density drops sharply is identified as the optical occlusion boundary; according to the relationship between the temperature gradient direction and the change of the radar point cloud density, an interpolation algorithm is selected to reconstruct the three-dimensional contour within the occlusion boundary: when the high temperature extends towards the direction of sparse point cloud, the depression repair algorithm is used to fill the contour; when the high temperature extends towards the direction of dense point cloud, the protrusion repair algorithm is used to correct the contour.

[0027] Exemplarily, for the extraction of the feature point set of the millimeter-wave radar point cloud, the Harris3D corner detection algorithm is used, and the top 20% of the points with the response value are extracted as feature points. The centroid is calculated as the average value of the feature point coordinates. The centroid of the infrared temperature anomaly region is obtained by the weighted average of all pixel coordinates within the region (the weight is the temperature value). The three-dimensional space distance is calculated as the square root of the sum of the squares of the differences of each coordinate component. When the distance < 0.5 meters, it is marked as a credible region. The texture continuity detection of the visible light image is performed by calculating the variance of the gradient amplitude of the pixels within the bounding box. When the variance is less than the threshold (such as 15), it is determined that the texture is continuous. The sharp drop in the radar point cloud density is defined as a reduction of more than 60% in the number of point clouds per unit volume (compared with the adjacent region), and the occlusion boundary is determined accordingly. The depression repair algorithm uses surface reconstruction based on the Poisson equation to fit the surface of the missing region with the occlusion boundary as the constraint condition. The protrusion repair algorithm uses triangular mesh subdivision, adds vertices at the occlusion boundary, and expands the mesh vertices outward according to the temperature gradient direction. The expansion distance is 0.1 times the gradient amplitude. The time interval of the millimeter-wave radar point cloud data is 20 ms, which is synchronized with the camera frame rate through timestamp interpolation to ensure the consistency of the spatial position.

[0028] In this embodiment, the system first screens the credible regions through the spatial correlation between the feature point set and the temperature anomaly region, locates the occlusion boundary within the credible regions by combining the visible light texture and the change in the radar point cloud density, and selects an appropriate interpolation algorithm to reconstruct the contour according to the temperature gradient direction. This method effectively solves the problem of the missing three-dimensional model caused by optical occlusion under complex lighting conditions. By fusing the temperature features and geometric features, it improves the reconstruction accuracy of the surface details of the obstacle, providing key data support for accurately evaluating the occupancy degree of the obstacle on the lane.

[0029] In another embodiment, the processing before calling the vehicle positioning module in the above method further includes: judging the material property of the obstacle according to the reflection intensity data of the millimeter-wave radar, and setting the priority of the metal obstacle as the highest; when the bandwidth of the cellular network is lower than the preset value, starting the data compression engine: using differential coding compression for the three-dimensional contour data of consecutive frames, and retaining the original data of the highest temperature region while compressing the background region of the infrared image; dynamically selecting the target server cluster for uploading according to the disaster type database, sending the landslide data to the geological disaster monitoring server, and sending the debris flow data to the hydrological monitoring server.

[0030] Exemplarily, the material judgment of the millimeter-wave radar reflection intensity data can be achieved through a preset threshold. The reflection intensity threshold for metal materials is set above -10 dBm, and for non-metals (such as trees and soil), it is set below -10 dBm. The radar model selected is Bosch MRRevo2, and its reflection intensity measurement accuracy is ±3 dB. In the data compression engine, differential coding compression is used for two consecutive frames of three-dimensional contour data, and only the differences in vertex coordinates are stored, with a compression ratio of up to 5:1. Infrared image compression uses ROI (Region of Interest) technology, automatically identifying the top 20% of the areas with the highest temperature as the original data to be retained, and the background area uses the JPEG compression algorithm with a compression quality factor set to 30. The preset value of the cellular network bandwidth is set to 100 kbps, and the bandwidth status is judged by real-time monitoring of the signal strength and error rate. The disaster type database contains 10 common disasters (such as landslides, mudslides, earthquake ruins, etc.), and each disaster type corresponds to a dedicated server cluster address. The vehicle positioning module selects the u-blox NEO-M8N GPS module with a positioning accuracy of 2.5 meters, which is assembled at the antenna position on the vehicle roof.

[0031] In this embodiment, the system first distinguishes the obstacle material according to the radar reflection intensity, and preferentially processes high-risk metal obstacles (such as collapsed steel frames and scattered vehicle parts) to ensure real-time response to emergency data. When the network bandwidth is insufficient, differential coding and ROI compression technologies are used to reduce the data transmission volume, ensuring the effective upload of key information (such as the highest temperature area and three-dimensional contour changes). Combining with the disaster type database, the server is dynamically selected to achieve the directional processing of professional data and improve the cloud analysis efficiency. This process optimizes the data processing priority and transmission strategy, ensures reliable communication in the complex network environment after the disaster, and provides timely and effective data support for subsequent risk assessment.

[0032] In another embodiment, the spatial matching of the cloud server in the above method specifically includes: extracting the historical deformation parameters of the road ahead of the vehicle from the digital elevation model of the disaster area, and calculating the geological settlement trend coefficient of the projection position of the current obstacle; establishing a virtual lane expansion model, superimposing the influence amount of the geological settlement trend coefficient on the lane width reference value to generate a dynamic lane width reference value; performing stress deformation simulation on the three-dimensional contour of the obstacle along the driving direction to predict the maximum expansion value of the projected area within the next 10 seconds; using the real-time ratio of the dynamic lane width reference value to the maximum expansion value as the spatial matching index.

[0033] Exemplarily, the historical deformation parameters of the disaster area digital elevation model are extracted from the surface elevation change data in the past 72 hours, with a time resolution of 1 hour. The data source is UAV aerial survey or ground settlement monitoring stations. When calculating the geological settlement trend coefficient, a dynamic monitoring grid with a radius of 5 meters is established centered on the projection position of the obstacle, and the deformation rate (unit: mm / h) of each coordinate point within the grid is statistically calculated. If the deformation rate of a certain point exceeds 10 mm / h for 3 consecutive hours and the direction is the same (uplift or settlement), it is marked as an active deformation point. The trend coefficient formula is: coefficient = reference value (1.0) + (proportion of active deformation points × 0.5 + average deformation acceleration × 0.3). When the proportion of active deformation points exceeds 40% and the acceleration is positive, the coefficient is adjusted up to 1.2 - 1.5 times; when the proportion is less than 10% and the acceleration is negative, the coefficient is adjusted down to 0.6 - 0.8 times. In the virtual lane expansion model, the reference value of the lane width is taken as the national standard value of 3.5 meters, and the dynamic width after superimposing the trend coefficient is 3.5 × coefficient. For stress deformation simulation, different models are adopted according to the obstacle material (rigid / flexible): for rigid materials, the thermal expansion effect is considered, and for every 10°C increase in temperature, the axial extension margin increases by 0.05 meters; for flexible materials, the softening effect of rainwater is considered, and for every 10 mm increase in precipitation, the expansion envelope range expands by 0.1 meters. The simulation time step is set to 0.1 seconds, and a total of 100 steps (the next 10 seconds) are calculated.

[0034] In this embodiment, the cloud server identifies the geologically active area by analyzing the historical elevation data, dynamically adjusts the lane width to adapt to the settlement or uplift trend, and predicts the deformation expansion range in combination with the obstacle material characteristics at the same time. This method combines the road geological conditions with the dynamic changes of obstacles, making the spatial matching index closer to the actual road conditions after the disaster, avoiding the limitations of the traditional fixed lane model, providing a more accurate basis for judging lane occupancy for risk assessment, and improving the scientificity and reliability of the alarm decision-making.

[0035] In another embodiment, the steps of extracting historical deformation parameters in the above method specifically include: extracting the surface elevation change data of the road in front of the vehicle per hour in the recent 72 hours from the disaster area digital elevation model to construct a time series deformation map; identifying the periodic uplift or settlement area in the deformation map. If the deformation rate of the same coordinate point exceeds the threshold for 3 consecutive hours and the change direction is the same, it is marked as an active deformation point; a dynamic monitoring grid with a radius of 5 meters is established centered on the projection position of the obstacle, and the proportion of active deformation points and the average deformation acceleration within the grid are statistically calculated; the geological settlement trend coefficient is dynamically calculated according to the weighted value of the proportion of active deformation points and the average deformation acceleration.

[0036] Exemplarily, the time series map of surface elevation change data has Coordinated Universal Time (UTC) on the horizontal axis and elevation change amount (unit: mm) on the vertical axis, with a resolution of 1 meter × 1 meter grid. The deformation rate threshold is set at 10 mm / h, and the rate is calculated from the elevation difference between two adjacent hours. Consistent direction means that there are positive changes (uplifts) or negative changes (settlements) for three consecutive hours. The dynamic monitoring grid is a square grid with a side length of 1 meter, and there are 25 grid points in total (within a radius of 5 meters). The proportion of active deformation points is the ratio of the number of marked points in the grid to the total number of points. The average deformation acceleration is obtained by fitting the deformation rate data for three hours with a quadratic polynomial and calculating the derivative to get the acceleration value (unit: mm / h²). In the weighted calculation of the geological settlement trend coefficient, the proportion weight is 0.6, the acceleration weight is 0.4, and the reference value is 1.0. The specific formula is: coefficient = 1.0 + 0.6×(proportion - 0.2) + 0.4×(acceleration / 50), where when the proportion exceeds 0.4, an upward adjustment mechanism is triggered, and when it is lower than 0.1, a downward adjustment mechanism is triggered to ensure that the coefficient is dynamically adjusted within the range of 0.6 - 1.5. The digital elevation model data is stored in a cloud database in the GeoTIFF format, including Universal Transverse Mercator (UTM) projection coordinates and elevation values, with an update frequency of once per hour.

[0037] In this embodiment, the system accurately identifies geological active areas by constructing a deformation map with a long time series, and improves the accuracy of settlement trend judgment by using dynamic grid statistical analysis. This process fully considers the spatio-temporal variation characteristics of the road after the disaster, avoids the one-sidedness of data at a single time point, enables the geological settlement trend coefficient to reflect the stability of the current road section in real time, provides reliable basic parameters for the virtual lane expansion model, and further optimizes the calculation accuracy of the obstacle projection area, enhancing the dynamic adaptability of risk assessment.

[0038] In another embodiment, the step of predicting the maximum expansion value of the projected area in the above method specifically includes: identifying the surface texture features of the obstacle through an in-vehicle multi-spectral camera, classifying it as a rigid material or a flexible material in combination with the corrected three-dimensional contour data; applying a virtual wind resistance load in the driving direction, and the wind resistance load value is dynamically calculated according to the ratio of the vehicle's current speed to the obstacle height; if it is a rigid material, monitoring the infrared temperature change rate at the edge of the obstacle base, and activating the thermal expansion compensation model when the rate exceeds 2 °C / second, adding an axial extension margin in the projected area calculation; if it is a flexible material, retrieving the current precipitation data and simulating the rainwater infiltration and softening effect, calculating the envelope range of deformation expansion within the next 10 seconds based on the material porosity parameter; taking the maximum superposition value of the thermal expansion compensation margin and the softening effect envelope range as the maximum expansion value.

[0039] Exemplarily, the classification of obstacle materials is achieved through texture analysis of visible light images. Rigid materials (such as rocks and metals) have clear texture edges and high gradient amplitudes, while flexible materials (such as soil and vegetation) have blurred textures and low gradient amplitudes. The classification model uses a support vector machine (SVM), and the training samples include 5000 material images of disaster scenes. The formula for calculating the virtual wind resistance load is: Load = 0.5 × air density × vehicle speed² × obstacle frontal area × drag coefficient, where the air density is taken as 1.2 kg / m³, the drag coefficient for rigid materials is set to 0.8, and the drag coefficient for flexible materials is set to 1.2. The monitoring of the infrared temperature change rate takes the average temperature change in a 3×3 pixel area at the base edge. With a time interval of 1 second, the compensation model is activated when it exceeds 2 °C / second continuously for 3 times, and the extension margin is 0.05 m / 10 °C temperature rise. The precipitation data is obtained through an in-vehicle meteorological sensor with an accuracy of 0.1 mm. The softening effect calculation assumes that for every 10% increase in porosity, the expansion coefficient increases by 0.08, and the envelope range is the expansion coefficient × time (seconds) by which the coordinates of each vertex of the current contour expand along the normal direction. The multispectral camera selects FLIR Boson 320, which is assembled above the front windshield and has a time synchronization error of less than 10 ms with the millimeter-wave radar (Bosch MRRevo2).

[0040] In this embodiment, the system, through material classification and multi-physical field coupling analysis, respectively considers the effects of thermal expansion and rain softening on the deformation of obstacles, and dynamically adjusts the expansion prediction model in combination with the wind resistance load. This process conducts refined modeling for the physical characteristics of different materials, avoids the prediction deviation of a unified model, makes the maximum expansion value closer to the actual deformation trend, and thus improves the accuracy of the projected area calculation, providing key data support for the subsequent accurate evaluation of the lane occupancy ratio and the reasonable planning of the avoidance path.

[0041] In another embodiment, the steps of generating an alarm instruction in the above method include: retrieving the avoidance trajectory data of all vehicles in the current section in the past 24 hours from the cloud server to construct a probability density distribution heat map; screening candidate paths that meet the vehicle kinematic constraints in the heat map according to the wheelbase parameter of the vehicle and the lateral extension length of the obstacle; preferentially selecting the path with the highest coincidence degree with the geologically stable zone in the digital elevation model of the disaster area among the candidate paths; decomposing the selected path into a longitudinal deceleration curve and a lateral offset instruction sequence, where the lateral offset instruction includes a steering angle safety margin value calculated in real time based on the tire grip.

[0042] Exemplarily, the avoidance trajectory data is stored in a cloud database in the format of (timestamp, longitude, latitude, steering wheel angle, vehicle speed). The heat map is constructed using kernel density estimation (KDE), with the bandwidth parameter set to 5 meters, generating a probability distribution grid with a resolution of 0.5 meters. The darker the color, the higher the passing probability. The vehicle kinematic constraint conditions are as follows: the minimum turning radius of the candidate path ≥ the wheelbase of the vehicle × 1.5 (the wheelbase parameter is obtained through the OBD interface, with a typical value of 2.7 meters), and the lateral offset ≤ 3.5 meters (lane width). The geological stable zone data comes from a digital elevation model, marking areas where the included angle between the strike of the rock formation fracture surface and the driving direction ≥ 30 degrees and the deformation rate in the past 72 hours < 5 mm / h. The overlap degree is calculated as the ratio of the overlapping length of the path and the stable zone to the total path length, and paths with an overlap degree > 60% are preferentially selected. The longitudinal deceleration curve is fitted with a quadratic function, with the initial speed being the current vehicle speed and the target speed being 0, and the braking distance meeting the requirements of GB21670-2008; the steering angle safety margin of the lateral offset command is calculated based on the tire grip formula: margin = 1 - (lateral acceleration / 0.8g), where the lateral acceleration is measured in real time by the vehicle inertial sensor, and 0.8g is the tire limit grip threshold.

[0043] In this embodiment, the system analyzes the high-frequency safe paths through the historical trajectory heat map, combines the screening of the geological stable zone and the vehicle kinematic constraints, and generates an avoidance path that conforms to the actual working conditions. This process makes full use of the group driving experience and geological stability data, avoids the limitations of a single algorithm, and at the same time dynamically adjusts the steering margin through real-time grip calculation to ensure that the path planning is both safe and in line with the vehicle's handling characteristics, providing reliable intervention instructions for the driver or the autonomous driving system, and effectively reducing the driving risk under complex road conditions after a disaster.

[0044] In another embodiment, the step of screening candidate paths in the above method specifically includes: extracting the strike data of the rock formation fracture surface from the digital elevation model of the disaster area, and marking the fracture zone with an included angle less than 30 degrees with the driving direction as the potential instability area; obtaining the ground vibration monitoring data of the current section in the past 2 hours, and if a certain candidate path passes through a grid cell with a vibration amplitude exceeding 0.3g and a duration exceeding 10 seconds, the geological stability coefficient of this path is downgraded; calculating the surface bearing capacity margin of each candidate path according to the vehicle full-load mass parameter, and eliminating the paths with a margin value lower than the safety threshold; weighting and sorting the remaining paths according to the fracture zone avoidance rate and the surface bearing capacity margin, and selecting the path with the highest comprehensive score.

[0045] Exemplarily, the strike data of the rock formation fracture surface is from the geological exploration database and stored as a vector file, including the location of the fracture zone and the strike angle (angle with the due north direction). The area with an angle with the driving direction less than 30 degrees is marked as a high-risk area. The ground vibration monitoring data is collected by acceleration sensors deployed on the road section, with a frequency of 100 Hz. The area where the vibration amplitude exceeds 0.3 g and lasts for more than 10 seconds is defined as an active vibration area and marked to the digital elevation model through grid division (10 m × 10 m). The surface bearing capacity margin is calculated as: Margin = (Road surface design bearing capacity - Vehicle full load mass / Contact area) / Road surface design bearing capacity, where the vehicle full load mass is obtained through in-vehicle weighing sensors, and the contact area is assumed to be the sum of the grounding areas of 4 tires (typical value 0.12 square meters), and the safety threshold is set to 0.2. The fracture zone avoidance rate is the ratio of the length of the path avoiding potential instability areas to the total path length. In the weighted ranking, the weight of the avoidance rate is 0.6, and the weight of the bearing capacity margin is 0.4. The comprehensive score = 0.6 × Avoidance rate + 0.4 × Margin. The top 3 paths with the highest scores are selected for further optimization. The digital elevation model, vibration data, and fracture zone data are integrated through a unified coordinate system (WGS84) to ensure the spatial matching accuracy.

[0046] In this embodiment, the system evaluates the path safety from three dimensions of the fracture zone strike, ground vibration, and surface bearing capacity through multi-source geological data fusion, and avoids entering potential instability areas. This process quantifies the geological risk factors and realizes the comprehensive optimization of the path through weighted ranking, ensuring that the avoidance path not only avoids immediate obstacles but also can avoid long-term geological disaster risks, improving the comprehensiveness and reliability of the post-disaster driving path planning and providing a safer driving plan for vehicles.

[0047] In another embodiment, the steps of decomposing the path instruction in the above method specifically include: obtaining the real-time tire pressure monitoring value, road surface infrared temperature value, and road surface roughness index detected by the millimeter-wave radar; when the tire pressure is 15% lower than the standard value and the road surface temperature is higher than 50 °C, activate the high-temperature soft tire compensation mode and increase the basic safety margin by 20%; if the road surface roughness index drops by more than 40% continuously for 5 frames, it is determined as a slippery road surface, and a sine damping factor is injected into the lateral offset instruction to suppress steering jitter; dynamically adjust the safety margin according to the real-time yaw angular velocity feedback of the vehicle: when the yaw angular velocity exceeds 0.5 rad / s, an additional 5% margin is added for every 0.1 rad / s increase; when the yaw angular velocity is lower than 0.2 rad / s for 3 seconds, the reference margin value is restored.

[0048] Exemplarily, the tire pressure monitoring value is obtained through the TPMS (Tire Pressure Monitoring System), and the standard value is set according to the vehicle manual (typical value 2.5 bar), with the accuracy of the tire pressure sensor being ±0.1 bar. The road surface infrared temperature value is collected by an in-vehicle infrared camera (FLIRT400) with a resolution of 320×240 and a temperature accuracy of ±2°C. The road surface roughness index is calculated from the signal-to-noise ratio of the millimeter-wave radar echo signal, and the formula is: roughness = 10×log10 (signal power / noise power), with the index range from 0 to 100, and a higher value indicating better roughness. The triggering condition for the high-temperature soft tire compensation mode is: tire pressure < 2.125 bar and temperature > 50°C, with the basic safety margin reference value being 10% and the increased value being 12%. The determination condition for a slippery road surface is: the roughness index continuously drops by > 40% for 5 frames (each frame is 0.1 second), the sine damping factor frequency is set to 2 Hz, and the amplitude is 0.1 rad to suppress the high-frequency jitter of the steering system. The yaw angular velocity is obtained by the inertial measurement unit (IMU) with an accuracy of ±0.05 rad / s, and the dynamic adjustment rule is: when the angular velocity > 0.5 rad / s, the margin = reference value + 5%×(angular velocity - 0.5) / 0.1, with the maximum increase up to 30%; when the angular velocity < 0.2 rad / s and lasts for 3 seconds, the margin returns to 10% of the reference value.

[0049] In this embodiment, the system monitors the tire status, road surface temperature, and roughness in real time, dynamically adjusts the steering safety margin for complex working conditions such as high temperature and slippery roads, suppresses steering jitter through the damping factor, and realizes the adaptive adjustment of the margin by combining the yaw angular velocity feedback. This process fully considers the interaction characteristics between the vehicle and the road surface, avoids the limitations of fixed-parameter control, makes the lateral offset command more in line with the real-time driving conditions, improves the handling stability and path tracking accuracy of the vehicle on complex road surfaces after disasters, and effectively reduces the risk of skidding or loss of control caused by changes in road surface conditions.

[0050] Although the embodiments of the present invention have been disclosed as above, it is not limited to only the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to the specific details and the illustrated and described examples here.

Claims

1. Intelligent alarm method for driving intervention after disasters, characterized in that, Including: S1: Collect visible light images and infrared thermal imaging images within the range of 0 - 150 meters in front of the vehicle through an in-vehicle multi-spectral camera, and simultaneously scan three-dimensional spatial coordinate data within the range of 0 - 150 meters in front through an in-vehicle millimeter-wave radar; S2: Input the visible light image into a convolutional neural network model for road obstacle type recognition, and at the same time perform temperature gradient analysis on the infrared thermal imaging image, and generate three-dimensional contour data of the obstacle in combination with the three-dimensional spatial coordinate data; S3: When the height of the obstacle exceeds the height threshold and the distance from the vehicle is less than the distance threshold, call the vehicle positioning module to obtain the current GPS coordinates, and upload a feature data set including the GPS coordinates, the three-dimensional contour data, and the obstacle type to the cloud server through the cellular network; S4: Based on the three-dimensional spatial coordinate data, the cloud server performs spatial matching between the feature data set and a pre-stored digital elevation model of the disaster area, and calculates the ratio of the projected area of the obstacle on the cross-section of the current lane to the lane width; S5: When the ratio exceeds the ratio threshold, generate an alarm instruction including the three-dimensional coordinates of the obstacle, the obstacle type, and avoidance path parameters, and send the alarm instruction to the vehicle through the cellular network.

2. The intelligent alarm method for driving intervention after a disaster according to claim 1, wherein S2 further includes: S21: Based on the temperature values of each pixel point in the infrared thermal imaging image, divide multiple temperature intervals and mark them as different color channels; S22: Identify continuous temperature anomaly regions. If the temperature difference between a certain region and its surroundings exceeds the dynamic threshold and lasts for more than 3 seconds, it is determined as a residue of thermal radiation disaster or a low-temperature icing area; S23: Perform spatial overlay of the obstacle bounding box recognized by the convolutional neural network in the visible light image and the temperature anomaly region. When the overlap degree exceeds 80%, activate the multi-modal verification mechanism, and reconstruct the surface deformation characteristics of the obstacle through millimeter-wave radar point cloud data; S24: According to the corresponding relationship between the surface deformation characteristics and the temperature distribution, correct the geometric distortion caused by optical occlusion in the three-dimensional contour data.

3. The intelligent alarm method for driving intervention after a disaster according to claim 2, wherein Locate the spatial feature point set of the obstacle through millimeter-wave radar point cloud data. If the distance between the temperature anomaly region in the infrared thermal imaging and the centroid of the feature point set is less than 0.5 meters, it is marked as a credible thermal radiation region; Within the credible thermal radiation region, identify the boundary where the texture of the visible light image is continuous but the radar point cloud density drops suddenly as the optical occlusion boundary; According to the relationship between the temperature gradient direction and the change in radar point cloud density, select an interpolation algorithm to reconstruct the three-dimensional contour within the occlusion boundary: when the high temperature extends towards the direction of sparse point cloud, use the depression repair algorithm to fill the contour; when the high temperature extends towards the direction of dense point cloud, use the protrusion repair algorithm to correct the contour.

4. The intelligent alarm method for driving intervention after a disaster according to claim 1, wherein S3 further includes: S31: Before calling the vehicle positioning module, judge the material property of the obstacle according to the reflection intensity data of the millimeter-wave radar, and set the priority of metal obstacles to the highest; S32: When the cellular network bandwidth is lower than the preset value, start the data compression engine: use differential coding compression for the three-dimensional contour data of consecutive frames, and retain the original data of the highest temperature region in the infrared image while compressing the background region; S33: Dynamically select the target server cluster for uploading according to the disaster type database. Landslide data is sent directly to the geological disaster monitoring server, and debris flow data is sent to the hydrological monitoring server.

5. The intelligent alarm method for driving intervention after a disaster according to claim 1, characterized in that, The specific steps of S4 include: S41: Extract the historical deformation parameters of the road ahead of the vehicle from the digital elevation model of the disaster area, and calculate the geological settlement trend coefficient of the current obstacle projection position; S42: Establish a virtual lane expansion model, superimpose the influence amount of the geological settlement trend coefficient on the reference value of the lane width, and generate a dynamic lane width reference value; S43: Conduct stress deformation simulation on the three-dimensional contour of the obstacle along the driving direction, and predict the maximum expansion value of the projected area within the next 10 seconds; S44: Use the real-time ratio of the dynamic lane width reference value to the maximum expansion value as the spatial matching index.

6. The intelligent alarm method for driving intervention after a disaster according to claim 5, wherein The specific steps of S41 include: S411: Extract the surface elevation change data of the road ahead of the vehicle per hour in the past 72 hours from the digital elevation model of the disaster area, and construct a time series deformation map; S412: Identify the periodically uplifted or subsided areas in the deformation map. If the deformation rate of the same coordinate point exceeds the threshold for 3 consecutive hours and the change directions are the same, it is marked as an active deformation point; S413: Establish a dynamic monitoring grid with a radius of 5 meters centered on the obstacle projection position, and count the proportion of active deformation points and the average deformation acceleration within the grid; S414: Dynamically calculate the geological settlement trend coefficient according to the weighted value of the proportion of active deformation points and the average deformation acceleration. Among them: when the proportion of active deformation points exceeds 40% and the acceleration is positive, the coefficient value is adjusted up to 1.2 - 1.5 times the reference value; when the proportion of active deformation points is lower than 10% and the acceleration is negative, the coefficient value is adjusted down to 0.6 - 0.8 times the reference value.

7. The intelligent alarm method for driving intervention after a disaster according to claim 6, wherein, The specific steps of S43 include: S431: Identify the surface texture features of the obstacle through an in-vehicle multi-spectral camera, and classify it as a rigid material or a flexible material in combination with the corrected three-dimensional contour data in claim 2; S432: Apply a virtual wind resistance load in the driving direction, and the value of the wind resistance load is dynamically calculated according to the ratio of the real-time speed of the vehicle to the height of the obstacle; S433: If it is a rigid material, monitor the infrared temperature change rate at the edge of the obstacle base. When the rate exceeds 2°C / second, activate the thermal expansion compensation model and add an axial extension margin to the projected area calculation; S434: If it is a flexible material, retrieve the current precipitation data and simulate the rainwater penetration and softening effect, and calculate the envelope range of the deformation expansion within the next 10 seconds based on the material porosity parameter; S435: Take the maximum superposition value of the thermal expansion compensation margin and the softening effect envelope range as the maximum expansion value.

8. The intelligent alarm method for driving intervention after a disaster according to claim 1, wherein, The specific steps of S5 include: S51: Retrieve the avoidance trajectory data of all vehicles on the current section in the past 24 hours from the cloud server, and construct a probability density distribution heat map; S52: Screen the candidate paths that meet the vehicle kinematic constraints in the heat map according to the wheelbase parameter of the vehicle and the lateral extension length of the obstacle; S53: Prioritize the selection of the path with the highest coincidence degree with the geological stable zone in the digital elevation model of the disaster area among the candidate paths; S54: Decompose the selected path into a longitudinal deceleration curve and a sequence of lateral offset instructions, where the lateral offset instructions include a steering angle safety margin value calculated in real time based on tire grip.

9. The intelligent alarm method for driving intervention after a disaster according to claim 8, wherein, The S53 includes: S531: Extract the strike data of the rock fracture surface from the digital elevation model of the disaster area, and mark the fracture zone with an angle less than 30 degrees with the driving direction as the potential instability area; S532: Obtain the ground vibration monitoring data of the current section in the past 2 hours. If a candidate path passes through a grid cell with a vibration amplitude exceeding 0.3g and a duration exceeding 10 seconds, downgrade the geological stability coefficient of this path; S533: Calculate the surface bearing capacity margin of each candidate path according to the vehicle full-load mass parameter, and eliminate the paths with a margin value lower than the safety threshold; S534: Weight and sort the remaining paths according to the fracture zone avoidance rate and the surface bearing capacity margin, and select the path with the highest comprehensive score.

10. The intelligent alarm method for driving intervention after a disaster according to claim 9, wherein The S54 includes: S541: Obtain the tire pressure monitoring value, the road surface infrared temperature value and the road surface roughness index detected by the millimeter wave radar in real time; S542: When the tire pressure is 15% lower than the standard value and the road surface temperature is higher than 50 °C, activate the high-temperature soft tire compensation mode and increase the basic safety margin by 20%; S543: If the road surface roughness index drops by more than 40% continuously for 5 frames, it is determined as a slippery road surface, and inject a sine damping factor into the lateral offset instruction to suppress steering jitter; S544: Dynamically adjust the safety margin according to the vehicle's real-time yaw angular velocity feedback: when the yaw angular velocity exceeds 0.5 rad / s, an additional 5% margin is added for every 0.1 rad / s increase; when the yaw angular velocity is lower than 0.2 rad / s for 3 seconds, restore the reference margin value.

Images