A dangerous goods road transportation remote electronic escort intelligent supervision system and method

Abstract

The application discloses a dangerous goods road transportation remote electronic escort intelligent supervision system and method, and belongs to the technical field of Internet of Things operating systems. The system comprises a data acquisition and fusion module, which acquires multi-source data of dangerous goods transportation vehicles in real time, and generates standardized fusion data flow; a transportation risk prediction and evaluation module, which generates a predictive risk map according to the standardized fusion data flow; a path planning module, which generates a risk avoidance path set based on the predictive risk map, and outputs a recommended path scheme; and an execution module, which generates graded early warning instructions and path adjustment suggestions according to the predictive risk map and the recommended path scheme, and simultaneously tracks the actual driving track and risk changes of the dangerous goods transportation vehicles in real time. The method comprises steps T1-T4. The application fuses historical and real-time data through the transportation risk prediction and evaluation module, constructs a predictive risk map, and realizes quantitative prediction and visual early warning of future accident risks of road sections.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) operating system technology, and in particular to a remote electronic escort intelligent supervision system and method for the road transport of dangerous goods. Background Technology

[0002] Dangerous goods road transport refers to the process of transferring chemicals or articles with explosive, flammable, toxic, corrosive, or other properties via freight trucks on public road networks. Its core safety requirement lies in the effective prevention and control of risks throughout the entire process and in multiple dimensions.

[0003] The working principle of existing technologies mainly relies on deploying various on-board sensors on transport vehicles to monitor key parameters such as vehicle speed, tank pressure, and axle temperature. The system pre-sets fixed alarm thresholds for each sensor. When the real-time reading of any sensor continuously or momentarily exceeds its corresponding threshold, a local audible and visual alarm is triggered and the alarm information is uploaded to the remote monitoring center, forming a passive response mechanism based on single-point threshold triggering.

[0004] However, the aforementioned existing technologies have significant drawbacks: their early warning logic is essentially a "post-event" response, lacking predictive early warning capabilities based on multi-source data fusion analysis. The system cannot comprehensively utilize multi-dimensional information such as real-time vehicle status, historical accident data, dynamic traffic information, and environmental changes to proactively assess and visualize the probability and level of accident risks in future periods along the transportation route. This makes it difficult for drivers and dispatch centers to promptly identify risk evolution trends and take proactive avoidance measures, resulting in a lag in safety decision-making. Therefore, there is an urgent need to provide a remote electronic escort intelligent supervision system and method for the road transportation of hazardous materials to solve these problems. Summary of the Invention

[0005] The technical problem this invention aims to solve is to overcome the fact that the early warning logic of the aforementioned existing technologies is essentially a "post-event" response, lacking predictive early warning capabilities based on multi-source data fusion analysis. The system cannot comprehensively utilize multi-dimensional information such as real-time vehicle status, historical accident data, dynamic traffic information, and environmental changes to proactively assess and visualize the probability and level of accident risks in future periods along the transportation route. This results in drivers and dispatch centers finding it difficult to promptly identify risk evolution trends and take proactive avoidance measures, leading to a lag in safety decision-making. Therefore, this invention provides a remote electronic escort intelligent supervision system and method for the road transportation of hazardous materials.

[0006] To solve the above-mentioned technical problems, one technical solution adopted by the present invention is: to provide a remote electronic escort intelligent supervision system for road transportation of dangerous goods, including a data acquisition and fusion module, a transportation risk prediction and assessment module, a route planning module and an execution module; The data acquisition and fusion module is used to collect multi-source data from dangerous goods transport vehicles in real time through the vehicle-mounted IoT terminal and the preset cloud interface, and to preprocess the multi-source data to generate a standardized fused data stream. The transportation risk prediction and assessment module, based on the standardized fusion data stream, predicts the probability and level of accidents on the route segment where the dangerous goods transport vehicle is located within a preset time period, and generates a predictive risk map. The route planning module, based on the predictive risk map and combined with road constraints and the transportation task requirements of hazardous materials transport vehicles, generates a risk avoidance route set, calculates the comprehensive safety score and expected travel time of each path in the risk avoidance route set, and outputs a recommended route plan. The execution module generates tiered early warning instructions and route adjustment suggestions based on the predictive risk map and the recommended route plan, and sends the tiered early warning instructions and route adjustment suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while simultaneously tracking the actual driving trajectory and risk changes of the hazardous materials transport vehicle in real time.

[0007] The present invention is further configured such that: the multi-source data in the data acquisition and fusion module includes status data of dangerous goods transport vehicles, environmental data, cargo data, and traffic data; The status data includes vehicle speed, axle temperature, and tank pressure; the environmental data includes real-time weather and road conditions; the cargo data includes hazardous material type and leak sensor readings; and the traffic data includes traffic congestion index and accident hotspot information.

[0008] The present invention is further configured such that the method for generating the predictive risk map in the transportation risk prediction and assessment module is as follows: S1. Based on the original planned route of the hazardous materials transport vehicle, map it to a preset two-dimensional path grid coordinate system to form a reference route. Divide the reference route into multiple continuous path grid units at preset distance intervals. Query a preset historical accident database to obtain historical road segment accident statistics with similar geographical features and traffic flow patterns to the current original planned route. Extract the historical accident frequency and severity data corresponding to each path grid unit. Combined with the inherent risk attributes of hazardous materials, calculate the static risk base of each path grid unit. The static risk base of all path grid units is spatially interpolated in the two-dimensional path grid coordinate system to generate an initial static risk field covering the current and future possible travel areas of the hazardous materials transport vehicle. S2. Based on the initial static risk field, preset risk impact factors are introduced for real-time correction. The risk impact factors include: vehicle operating status factors calculated based on real-time vehicle speed and key component temperature data of dangerous goods transport vehicles; environmental disturbance factors assessed based on real-time meteorological information and road surface adhesion coefficient; and cargo real-time risk factors determined based on the chemical stability of the dangerous goods transported by the dangerous goods transport vehicle and tank sealing sensor readings. A weighted superposition algorithm is used to fuse each of the risk impact factors with the static risk base of its corresponding path grid unit. After fusion calculation, the real-time dynamic risk value of each path grid unit is output to generate a dynamically updated risk field that reflects the current comprehensive risk status. S3. Compare the real-time dynamic risk value of each path grid cell in the dynamically updated risk field with a preset multi-level risk threshold. Based on the comparison result, mark the path grid cells whose real-time dynamic risk value falls into different threshold ranges of the multi-level risk threshold as different risk levels and distinguish them using different preset visual codes. Integrate the risk level markings of all path grid cells to generate a predictive risk map on the two-dimensional path grid coordinate system.

[0009] The present invention is further configured such that the steps for generating the risk avoidance path set in the path planning module are as follows: Q1. Based on the risk level of each path grid cell in the predictive risk map, identify continuous path segments in the baseline path whose risk level exceeds a preset safety threshold and mark them as high-risk path segments; in the two-dimensional path grid coordinate system, delete the high-risk path segments from the baseline path to form a main path containing a path interruption zone, wherein the path interruption zone is jointly defined by the starting deletion node and the ending deletion node left after the high-risk path segment is deleted; Q2. For each of the path interruption zones, in the two-dimensional path grid coordinate system, with the starting deletion node and the ending deletion node as spatial anchor points, an adjustment path framework is constructed to bypass the high-risk path segment; the adjustment path framework consists of a set of adjustment points pre-generated based on the geographical topological relationship in the vicinity of the benchmark path, and the adjustment points are automatically generated by analyzing the road network connectivity around the path interruption zone to form an adjustment point sequence. Q3. Based on the road constraints of each path within the preset range adjacent to the main path and the transportation requirements of the hazardous materials transport vehicle, optimize and adjust the sequence of adjustment points; dynamically distribute the optimized adjustment points to each road intersection, road segment midpoint, and road end point of the adjustment path framework to form a final adjustment point set; sequentially connect each final adjustment point in the final adjustment point set to generate a local risk avoidance path; seamlessly topologically stitch the local risk avoidance path with the main path to output a complete risk avoidance path, and repeat the above process to generate multiple local risk avoidance paths to form the risk avoidance path set.

[0010] The present invention is further configured such that the method for constructing the adjustment path framework in step Q2 is as follows: Q201. For each path interruption zone, in the two-dimensional path grid coordinate system, identify all road intersections and connecting road segments within a preset buffer radius around the starting deletion node and the ending deletion node, construct a local road network topology map, and extract key topological features from the local road network topology map, including the connectivity of road intersections, the traffic capacity level of road segments, and road network connectivity indicators, to form a set of road network structure features within the preset buffer radius around the path interruption zone. Q202. Calculate the topological importance score of each road intersection in the local road network topology map. The topological importance score is obtained by weighted summation of the number of connecting roads and the capacity level of the connecting roads at the road intersection. Sum the topological importance scores of all road intersections on the boundary of the path interruption zone and multiply them by a normalization coefficient based on the road network connectivity index to generate a reference value for the number of basic adjustment points. Round the reference value for the number of basic adjustment points up. The rounded reference value for the number of basic adjustment points is not lower than a preset minimum adjustment point threshold. Output the number of basic adjustment points. Q203. Based on the topological importance score of each road intersection in the local road network topology map surrounding the path interruption zone, the road intersections are classified and screened. The top N road intersections with the highest topological importance scores are determined as core adjustment points, which constitute the backbone nodes of the path detour. If the number of core adjustment points is insufficient, auxiliary adjustment points are selected from the remaining road intersections whose connectivity exceeds a preset connectivity threshold and whose traffic capacity level exceeds a preset level. All core adjustment points and auxiliary adjustment points are sorted according to their spatial topological relationship with the starting deletion node and the ending deletion node to form a final adjustment point sequence. The starting deletion node is used as the path starting point, and each point in the final adjustment point sequence is connected sequentially, with the ending deletion node as the endpoint, to generate an adjustment path framework.

[0011] The present invention is further configured such that: the road constraints in step Q3 include road grade restrictions, load restrictions, and traffic control rules; the transportation task requirements for dangerous goods transport vehicles include the expected arrival time window and the hazard level of the goods; The specific content of optimizing and adjusting the adjustment point sequence is as follows: Based on the adjustment point sequence, combined with the road grade restrictions, load restrictions, and traffic control rules, calculate the road traffic feasibility index for each adjustment point; simultaneously, based on the expected arrival time window and cargo hazard level, evaluate the time window matching degree and risk control priority of each adjustment point; input the road traffic feasibility index, the time window matching degree, and the risk control priority into a preset multi-objective optimization function, and generate a comprehensive adjustment coefficient for each adjustment point through weighted summation; based on the level of the comprehensive adjustment coefficient, rearrange the adjustment points in the adjustment point sequence, remove the adjustment points whose comprehensive adjustment coefficient is lower than a preset feasibility threshold, and output the optimized and adjusted adjustment point sequence.

[0012] The present invention is further configured such that the method for outputting the recommended path scheme in the path planning module is as follows: Q4. For each local risk avoidance path in the risk avoidance path set, extract its path geometric features and risk distribution features. The path geometric features include the total path length, the sum of turning angles, and the average radius of curvature. The risk distribution features are characterized by the average and maximum values ​​of the real-time dynamic risk values ​​of all the path grid cells traversed by the local risk avoidance path. Map the path geometric features and the risk distribution features to time cost coefficients and risk exposure coefficients, respectively. Combined with the expected arrival time window in the transportation task requirements of dangerous goods transport vehicles, calculate the time compliance score of the local risk avoidance path. Calculate the comprehensive safety score and expected travel time for each local risk avoidance path based on the time cost coefficient, risk exposure coefficient, and time compliance score. Q5. Standardize the comprehensive safety score and expected travel time of all local risk avoidance paths in the risk avoidance path set, and use the Pareto optimal ranking algorithm to sort each local risk avoidance path in multiple dimensions. Select the local risk avoidance paths with the top 30% comprehensive safety scores and expected travel times not exceeding a preset time threshold as the recommended path set. The recommended path set is the recommended path scheme.

[0013] The present invention is further configured such that the generation steps of the hierarchical early warning instruction and the path adjustment suggestion in the execution module are as follows: R1. Receive the predictive risk map and the recommended route plan, simultaneously acquire the real-time location coordinates and driving trajectory data of the dangerous goods transport vehicle, extract the current location of the dangerous goods transport vehicle and the real-time dynamic risk value of the adjacent path grid unit from the predictive risk map, and calculate the current comprehensive risk exposure value by combining the vehicle speed, axle temperature and tank pressure data collected by the real-time sensors of the dangerous goods transport vehicle. R2. Compare the current comprehensive risk exposure value with the preset multi-level risk thresholds. Based on the comparison results, map the threshold range into which the current comprehensive risk exposure value falls to a specific warning level and generate a preliminary warning instruction framework. At the same time, analyze the deviation between the actual driving trajectory of the dangerous goods transport vehicle and the recommended route plan. If the deviation exceeds the preset safety tolerance, trigger the route anomaly flag. R3. Based on the preliminary early warning instruction framework and the path anomaly marker, call the preset rule library to match the corresponding early warning content, response time and notification scope for each specific early warning level, and generate a structured hierarchical early warning instruction; simultaneously retrieve the avoidance strategy template corresponding to the current specific early warning level and the path anomaly marker from the preset path adjustment strategy library, and combine real-time traffic data and road constraints to parameterize and instantiate the avoidance strategy template to form a path adjustment suggestion.

[0014] A method for intelligent supervision of remote electronic escort of dangerous goods road transport includes the following steps: T1. Collect multi-source data of dangerous goods transport vehicles in real time through vehicle-mounted IoT terminals and preset cloud interfaces, and preprocess the multi-source data to generate a standardized fusion data stream; T2. Based on the standardized fused data stream, predict the probability and risk level of accidents on the route segment where the dangerous goods transport vehicle is located within a preset time period in the future, and generate a predictive risk map. T3. Based on the predictive risk map, combined with road constraints and the transportation task requirements of dangerous goods transport vehicles, generate a risk avoidance path set, calculate the comprehensive safety score and expected travel time of each path in the risk avoidance path set, and output a recommended route plan. T4. Based on the predictive risk map and the recommended route plan, generate graded early warning instructions and route adjustment suggestions, and send the graded early warning instructions and route adjustment suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while tracking the actual driving trajectory and risk changes of the dangerous goods transport vehicle in real time.

[0015] The beneficial effects of this invention are as follows: 1. This invention integrates multi-source heterogeneous data through a data acquisition and fusion module and generates a standardized fused data stream, realizing comprehensive and collaborative perception of vehicle status, environment, cargo and traffic information, laying a reliable data foundation for accurate risk assessment; 2. This invention integrates historical and real-time data through a transportation risk prediction and assessment module to construct a predictive risk map, thereby achieving quantitative prediction and visual early warning of future accident risks on road sections, transforming passive response into proactive prevention, and significantly improving the timeliness of risk response; 3. This invention, through the collaboration of the path planning module and the execution module, intelligently generates and executes risk avoidance paths and hierarchical early warnings based on prediction results, realizing closed-loop management from risk perception, decision-making to control, and ensuring the safety and efficiency of the transportation process. Attached Figure Description

[0016] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0018] Please see Figures 1-2 A remote electronic escort intelligent supervision system for road transport of dangerous goods includes a data acquisition and fusion module, a transport risk prediction and assessment module, a route planning module, and an execution module. The data acquisition and fusion module is used to collect multi-source data from dangerous goods transport vehicles in real time through the vehicle-mounted IoT terminal and the preset cloud interface, and to preprocess the multi-source data to generate a standardized fused data stream. The transportation risk prediction and assessment module, based on standardized fusion data streams, predicts the probability and level of accidents on the route segments where dangerous goods transport vehicles are located within a preset time period, and generates a predictive risk map. The route planning module, based on the predictive risk map and combined with road constraints and the transportation task requirements of dangerous goods transport vehicles, generates a risk avoidance route set, calculates the comprehensive safety score and expected travel time of each route in the risk avoidance route set, and outputs recommended route solutions. The execution module generates tiered early warning instructions and route adjustment suggestions based on the predictive risk map and recommended route schemes. It then sends these instructions and suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while simultaneously tracking the actual driving trajectory and risk changes of the hazardous materials transport vehicle in real time.

[0019] The data acquisition and fusion module, transportation risk prediction and assessment module, route planning module, and execution module together form the Internet of Things operating system.

[0020] The system's four modules work together. The data acquisition and fusion module enables real-time perception and standardized processing of multi-source information, providing a reliable data foundation for risk assessment. The transportation risk prediction and assessment module generates predictive risk maps based on data analysis, enabling early warning. The route planning module intelligently plans safe and efficient routes accordingly. The execution module ensures the timely issuance and closed-loop tracking of warnings and adjustment instructions, significantly improving the proactive safety, regulatory efficiency, and accident prevention capabilities of dangerous goods transportation.

[0021] The data acquisition and fusion module includes multi-source data such as the status data of dangerous goods transport vehicles, environmental data, cargo data, and traffic data. Status data includes vehicle speed, axle temperature, and tank pressure; environmental data includes real-time weather and road conditions; cargo data includes hazardous material types and leak sensor readings; and traffic data includes traffic congestion index and accident hotspot information.

[0022] Status data is acquired through in-vehicle IoT terminals (such as GPS, axle temperature sensors, and tank pressure sensors). Environmental data is acquired by calling meteorological service APIs and road condition detection systems through pre-defined cloud interfaces. Cargo data is acquired through in-vehicle hazardous materials identification electronic tags and tank sealing sensors. Traffic data is acquired through an accident hotspot and congestion index database integrated with a real-time traffic information platform.

[0023] The method for generating the predictive risk map in the transportation risk prediction and assessment module is as follows: S1. Based on the original planned route of the hazardous materials transport vehicle, map it to a preset two-dimensional path grid coordinate system to form a reference route. Divide the reference route into multiple continuous path grid units at preset distance intervals. Query the preset historical accident database to obtain historical road segment accident statistics with similar geographical features and traffic flow patterns to the current original planned route. Extract the historical accident frequency and severity data corresponding to each path grid unit. Combined with the inherent risk attributes of hazardous materials, calculate the static risk base of each path grid unit. Perform spatial interpolation calculation on the static risk base of all path grid units in the two-dimensional path grid coordinate system to generate an initial static risk field covering the current and future possible travel areas of the hazardous materials transport vehicle. The optimal value of the preset distance interval is based on the adaptive adjustment of road network density and vehicle operating speed. First, the average spacing of road intersections in the target area is statistically analyzed as the baseline distance. Then, combined with the average speed of dangerous goods transport vehicles and the timeliness requirements of risk prediction, the dynamic interval is calculated by multiplying the baseline distance and the speed adjustment coefficient. The final value is usually controlled within the range of 500 meters to 1000 meters to balance calculation accuracy and efficiency.

[0024] The pre-built historical accident database is constructed by integrating accident records from the transportation department, safety data reported by enterprises, and publicly available disaster information databases. This database stores the spatiotemporal coordinates, accident type, degree of casualties, and environmental factors of each accident according to a geographic grid index, and ensures consistency through data cleaning and normalization processing, providing standardized input for static risk base calculation.

[0025] The formula for calculating the static risk base of each path grid cell is as follows: multiply the historical accident frequency by the accident severity coefficient to obtain the original risk value, then multiply by the inherent risk weight coefficient of the hazardous materials corresponding to the grid cell, and finally introduce the environmental sensitivity correction factor for calibration to output the static risk base. The environmental sensitivity correction factor is obtained by integrating geospatial data, quantitatively assessing the vulnerability and importance of the surrounding environment of each path grid cell, and quantifying this assessment result into a coefficient used to adjust the basic risk value.

[0026] S2. Based on the initial static risk field, preset risk impact factors are introduced for real-time correction. These risk impact factors include: vehicle operating status factors calculated based on real-time vehicle speed and key component temperature data of hazardous materials transport vehicles; environmental disturbance factors assessed based on real-time meteorological information and road surface adhesion coefficient; and real-time cargo risk factors determined based on the chemical stability of the hazardous materials transported by the hazardous materials transport vehicles and tank sealing sensor readings. A weighted superposition algorithm is used to fuse each risk impact factor with the static risk base of its corresponding path grid unit. After the fusion calculation, the real-time dynamic risk value of each path grid unit is output, generating a dynamically updated risk field that reflects the current comprehensive risk status. Key components of vehicles transporting hazardous materials include axles and tanks.

[0027] The weights of dynamic risk impact factors are dynamically allocated based on the reliability of their current data and their contribution to risk. The vehicle operating status factor is calculated through the normalization of real-time vehicle speed and axle / tank temperature data. First, the deviation of vehicle speed and temperature from the safety threshold is calculated separately. Then, the two deviations are weighted and fused into a comprehensive status index. The environmental disturbance factor is calculated through joint analysis of real-time meteorological data and road surface adhesion coefficient, mapping wind speed, precipitation intensity, and road surface friction coefficient to environmental disturbance levels. The real-time cargo risk factor is calculated by multiplying the hazardous materials chemical stability index by the tank sealing sensor reading, combined with a leakage probability model. The fusion calculation of each risk impact factor and its corresponding path grid cell's static risk base adopts a dynamic weight allocation strategy. Each factor is multiplied by a weight and then added to the static risk base. The weights are dynamically adjusted based on the real-time reliability of the factor data, outputting a real-time dynamic risk value.

[0028] S3. The real-time dynamic risk value of each path grid cell in the dynamically updated risk field is compared with the preset multi-level risk thresholds. Based on the comparison results, path grid cells whose real-time dynamic risk values ​​fall into different threshold ranges of the multi-level risk thresholds are marked as different risk levels and distinguished using different preset visual codes. The risk level markings of all path grid cells are integrated to generate a predictive risk map on the two-dimensional path grid coordinate system, which can intuitively display the spatial distribution and level differences of risk in the future preset time period. The predictive risk map supports the simulation of risk evolution over time, providing a forward-looking risk avoidance basis for the path planning module.

[0029] Preset visual encoding includes color depth and pattern fill.

[0030] The steps for generating the risk avoidance path set in the path planning module are as follows: Q1. Based on the risk level of each path grid cell in the predictive risk map, identify continuous path segments in the baseline path whose risk level exceeds the preset safety threshold and mark them as high-risk path segments; in the two-dimensional path grid coordinate system, delete the high-risk path segments from the baseline path to form the main path containing the path interruption area. The path interruption area is jointly defined by the starting deletion node and the ending deletion node left after the high-risk path segment is deleted. The optimal value of the preset safety threshold is determined based on the quantile analysis of historical accident statistics. First, the 90th percentile value of the static risk base of all path grid units is calculated as the baseline threshold. Then, the threshold is fine-tuned by multiplying the level of dangerous goods (such as highly toxic or flammable) in the current transportation task by the risk level coefficient. The final threshold must ensure that the false alarm rate and the missed alarm rate of high-risk path segments are balanced.

[0031] When removing a high-risk path segment from the baseline path, it is necessary to consider whether the high-risk path segment is at a road intersection. If not, the high-risk path segment is extended to the intersection. The specific steps are as follows: First, determine whether the endpoint of the high-risk path segment is located at a road intersection. If not, extend it to both ends along the road network topology. The extension distance is determined by the Euclidean distance from the endpoint of the path segment to the nearest intersection. The extension direction must follow the road connectivity rules until the endpoint coincides with the intersection.

[0032] Q2. For each path interruption zone, in the two-dimensional path grid coordinate system, with the starting deletion node and the ending deletion node as spatial anchor points, construct an adjustment path framework to bypass the high-risk path segment; the adjustment path framework consists of a set of adjustment points pre-generated based on the geographical topological relationship in the vicinity of the benchmark path. The adjustment points are automatically generated by analyzing the road network connectivity around the path interruption zone, forming an adjustment point sequence. Neighboring area: The definition is based on the spatial topology of the path interruption zone. A buffer zone is constructed with the interruption zone as the center and the average distance between adjacent main roads in the road network hierarchy as the radius. All road nodes and road segments in this area are considered as neighboring areas.

[0033] Q3. Based on the road constraints of each path within the preset range adjacent to the main path, and the transportation requirements of the hazardous materials transport vehicles, optimize and adjust the sequence of adjustment points; dynamically distribute the optimized adjustment points to each road intersection, road segment midpoint, and road end point of the adjustment path framework to form a final adjustment point set; sequentially connect each final adjustment point in the final adjustment point set to generate a local risk avoidance path; seamlessly topologically stitch the local risk avoidance path with the main path to output a complete risk avoidance path, and repeat the above process to generate multiple local risk avoidance paths to form a risk avoidance path set.

[0034] The optimal value of the adjacent preset range is determined by road network density analysis. First, the total length of roads per unit area in the target area is calculated as the density index. Then, the density index is multiplied by the range scaling factor to obtain the dynamic range value, which is usually controlled within 1-2 kilometers outside the interruption zone.

[0035] The method for constructing the path framework in step Q2 is as follows: Q201. For each path interruption zone, in the two-dimensional path grid coordinate system, identify all road intersections and connecting road segments within the preset buffer radius around the starting and ending deletion nodes, construct a local road network topology map, extract key topological features from the local road network topology map, including the connectivity of road intersections, the traffic capacity level of road segments, and road network connectivity indicators, and form a set of road network structural features within the preset buffer radius around the path interruption zone. The optimal value of the preset buffer radius is set based on the correlation between the vehicle's turning radius and the road grade. It is based on three times the minimum turning radius of dangerous goods transport vehicles, and then multiplied by the grade coefficient according to the road grade (such as expressway or national highway). The final radius is generally set to 500 meters to 2 kilometers.

[0036] The steps for extracting the connectivity degree of road intersections, the capacity level of road segments, and the road network connectivity index are as follows: First, based on the constructed local road network topology map, traverse each road intersection node in the map and count the number of connecting road segments directly connected to that node. This number is the connectivity degree of the road intersection, reflecting the hub importance of the node in the local road network. Second, for each connecting road segment, read the preset road design parameters from its attribute data, including the number of lanes, design speed, and road technical grade. Input these parameters into a preset capacity mapping model, which outputs a representation of the road network capacity through a weighted fusion algorithm. The scalar value of the maximum service capacity of a road segment per unit time is the capacity level of that road segment. Then, by combining the connectivity of all road intersections and the capacity levels of all connected road segments, the overall connectivity efficiency of the local road network is calculated. This efficiency value is obtained by summing the products of the connectivity of all intersections and the capacity levels of their connected road segments, and then dividing by the product of the total number of intersections in the network and the benchmark of the average capacity level of the road segments. This calculation result is the road network connectivity index. Finally, the connectivity set, the capacity level set, and the road network connectivity index together constitute the set of road network structural characteristics around the path interruption area.

[0037] Q202. Calculate the topological importance score of each road intersection in the local road network topology map. The topological importance score is obtained by weighted summation of the number of connecting roads and the capacity level of the connecting roads at the road intersection. Sum the topological importance scores of all road intersections on the boundary of the path interruption zone and multiply them by a normalization coefficient based on the road network connectivity index to generate a reference value for the number of basic adjustment points. Round the reference value of the number of basic adjustment points up. The rounded reference value of the number of basic adjustment points is not lower than the preset minimum threshold for the number of adjustment points. Output the number of basic adjustment points. The method for calculating the topological importance score is as follows: the connectivity (i.e. the number of intersecting roads) of the road intersection and the capacity level of each connecting road are extracted. The weighted sum of the connectivity and capacity levels is used as the initial score, and then the historical traffic data of the intersection is introduced for normalization correction.

[0038] The normalization coefficient is calculated using the road network connectivity index, which is the ratio of the actual number of road connections to the theoretical maximum number of connections. After a logarithmic transformation, the ratio is mapped to the 0-1 interval as the coefficient.

[0039] The optimal value of the preset minimum number of adjustment points threshold is determined based on the path detour tolerance limit. First, based on vehicle navigation experience, the minimum detour coefficient is set to 1.5, and then multiplied by the straight-line distance of the path interruption zone to obtain the lower limit of the number of points. Usually, at least 3 adjustment points are required.

[0040] Q203. Based on the topological importance score of each road intersection in the local road network topology map surrounding the path interruption zone, road intersections are classified and screened. The top N road intersections with the highest topological importance scores are determined as core adjustment points (N is the number of basic adjustment points). The core adjustment points constitute the main nodes of the path detour. If the number of core adjustment points is insufficient, auxiliary adjustment points are selected from the remaining road intersections whose connectivity exceeds a preset connectivity threshold and whose traffic capacity level exceeds a preset level. All core adjustment points and auxiliary adjustment points are sorted according to their spatial topological relationship with the starting deletion node and the ending deletion node to form a final adjustment point sequence. Starting from the starting deletion node, each point in the final adjustment point sequence is connected sequentially, and ending at the ending deletion node to generate a preliminary adjustment path framework. The preliminary adjustment path framework is optimized for smoothness to ensure that the connection between adjacent nodes conforms to the vehicle steering geometry constraint. The final output is an adjustment path framework that can be used for the generation of subsequent local risk avoidance paths.

[0041] The optimal value of the preset connectivity threshold is 1.2 times the average connectivity of the road network. The average connectivity of all intersections in the region is calculated and then multiplied by a safety factor to ensure that the selected auxiliary adjustment points have sufficient redundant paths.

[0042] The preset level is set according to the traffic capacity classification in the road design standard. The traffic capacity level is divided into three levels, and the preset level is the second level or above to ensure that the road where the adjustment point is located meets the basic traffic needs of dangerous goods vehicles.

[0043] In step Q3, the road constraints include road grade restrictions, load limits, and traffic control rules; the transportation task requirements for dangerous goods transport vehicles include the expected arrival time window and the hazard level of the goods. The specific steps for optimizing the adjustment point sequence are as follows: Based on the adjustment point sequence, and in conjunction with road grade restrictions, load restrictions, and traffic control rules, calculate the road traffic feasibility index for each adjustment point; simultaneously, assess the time window matching degree and risk control priority of each adjustment point based on the expected arrival time window and the cargo hazard level; input the road traffic feasibility index, time window matching degree, and risk control priority into a preset multi-objective optimization function, and generate a comprehensive adjustment coefficient for each adjustment point through weighted summation; based on the level of the comprehensive adjustment coefficient, rearrange the adjustment points in the adjustment point sequence, remove adjustment points with a comprehensive adjustment coefficient lower than the preset feasibility threshold, and output the optimized adjustment point sequence.

[0044] The calculation steps for the road traffic feasibility index of the adjustment point are as follows: First, a binary compliance judgment result is generated based on road grade restrictions, load restrictions and traffic control rules. Then, the compliance result is quantified into a base score, and dynamic correction is performed in combination with real-time traffic flow data (such as congestion index) to output an index value in the range of 0-1.

[0045] The time window matching degree of each adjustment point is obtained by calculating the overlap between the expected arrival time and the task requirement time window, while the risk control priority is assessed based on the weighted product of the cargo hazard level and the sensitivity of the surrounding environment of the adjustment point (such as schools and hospitals).

[0046] The preset multi-objective optimization function adopts a linear weighted sum form, assigning weights to the road traffic feasibility index, time window matching degree, and risk control priority, respectively. The weight values ​​are dynamically configured based on the urgency and safety requirements of the transportation task.

[0047] The optimal value of the preset feasibility threshold is determined by back-calculating the success rate of historical paths. The distribution of the comprehensive adjustment coefficient of each adjustment point in the past successful paths is statistically analyzed, and the 10th percentile value is taken as the lower limit of the threshold, which is usually set to 0.6.

[0048] The recommended route scheme is output in the route planning module as follows: Q4. For each local risk avoidance path in the risk avoidance path set, extract its path geometric features and risk distribution features. The path geometric features include the total path length, the sum of turning angles, and the average radius of curvature. The risk distribution features are characterized by the average and maximum values ​​of the real-time dynamic risk values ​​of all path grid cells traversed by the local risk avoidance path. Map the path geometric features and risk distribution features to time cost coefficients and risk exposure coefficients, respectively. Combined with the expected arrival time window in the transportation task requirements of dangerous goods transport vehicles, calculate the time compliance score of the local risk avoidance path. Calculate the comprehensive safety score and expected travel time for each local risk avoidance path based on the time cost coefficient, risk exposure coefficient, and time compliance score. The specific steps for mapping path geometric features to time cost coefficients are as follows: divide the total path length by the standard vehicle speed to obtain the base time, then calculate the steering delay coefficient based on the sum of the turning angles and the average radius of curvature, and finally normalize the result by multiplying the two. The steps for mapping risk distribution features to risk exposure coefficients are as follows: weight and fuse the average and maximum risk values ​​of the path grid cells, and then multiply by the risk attenuation factor.

[0049] The calculation method for the time compliance score of the local risk avoidance path is as follows: the deviation between the expected travel time and the task requirement time window is transformed by a negative exponential transformation. The smaller the deviation, the higher the score. At the same time, a hard constraint of the time window is introduced as a boundary condition.

[0050] The calculation steps for the comprehensive safety score and expected travel time are as follows: The time cost coefficient, risk exposure coefficient, and time compliance score are input into the scoring model. The scoring model generates a safety score through a weighted product, and the expected travel time is directly output through a convolution operation between the path geometric features and the vehicle speed model. The scoring model can be either the Analytic Hierarchy Process (AHP) or the Fuzzy Comprehensive Evaluation Method.

[0051] Q5. Standardize the comprehensive safety score and expected travel time of all local risk avoidance paths in the risk avoidance path set, and use the Pareto optimal ranking algorithm to sort each local risk avoidance path in multiple dimensions. Select the local risk avoidance paths with the top 30% comprehensive safety scores and expected travel time not exceeding the preset time threshold as the recommended path set. The recommended path set is the recommended path scheme.

[0052] The optimal value of the preset time threshold is set based on the economic constraints of the transportation task. The upper limit threshold is set as 1.3 times the initial planned time of the task to ensure that the route planning result is within an acceptable time cost range.

[0053] Example A hazardous materials transportation company is transporting a batch of flammable liquids from a chemical plant in City A to a warehousing center in City B. Its vehicles are equipped with onboard IoT terminals that collect real-time status data such as vehicle speed (100 km / h), axle temperature (85 degrees Celsius), and tank pressure (0.5 MPa). Simultaneously, the cloud interface obtains real-time environmental data showing light rain and a road surface slippage coefficient of 0.3. Electronic tags identify the hazardous materials as Class III flammable liquids, and the leak sensor reading is 0. The traffic platform provides traffic data showing a current congestion index of 1.8 and two accident hotspots.

[0054] The data acquisition and fusion module performs timestamp alignment and anomaly cleaning on multi-source data to generate a standardized fused data stream. The transportation risk prediction and assessment module uses the original planned route as a baseline, dividing the path into grid units at 500-meter intervals. It queries the historical accident database to obtain a historical accident frequency of 0.05 times / month and a severity coefficient of 1.2 for a specific grid unit. Combining this with the inherent risk weight of 1.5 for flammable liquids, the static risk base is calculated to be 0.09. Introducing a vehicle operating status factor of 0.8, an environmental disturbance factor of 1.2, and a real-time cargo risk factor of 1.0, the module weighted and superimposed these factors to output a real-time dynamic risk value of 0.108. Comparing this to a preset threshold of 0.1, the unit is marked as high-risk, generating a predictive risk map.

[0055] The path planning module identifies a continuous 3-kilometer high-risk path segment and removes it from the baseline path to form a path interruption zone containing start and end nodes. With a buffer radius of 1000 meters, it extracts the connectivity of surrounding road intersections (3) and traffic capacity level (2), calculates a topological importance score of 6.5, and generates 4 basic adjustment points. The top 4 core adjustment points are selected, and combined with constraints such as a road weight limit of 30 tons and a task requirement of 8 hours of arrival, the feasibility index of the adjustment points is calculated to be 0.9, the time matching degree to be 0.85, and the risk priority to be 1.1. The weighted comprehensive adjustment coefficient is 0.84. After removing adjustment points with a coefficient lower than 0.6, a local avoidance path is generated.

[0056] The final recommended route, selected based on Pareto ranking, has a comprehensive safety score of 92 and an estimated travel time of 7.5 hours. The execution module issues warning commands and adjustment suggestions, and tracks vehicle trajectories in real time.

[0057] This embodiment achieves accurate identification and avoidance of high-risk road sections through multi-source data fusion and dynamic risk assessment; route optimization based on real-time constraints significantly improves transportation safety, with a comprehensive safety score of 92 points; closed-loop management throughout the entire process reduces the probability of accidents and shortens the expected travel time by 12%, achieving synergistic optimization of safety and efficiency.

[0058] The steps for generating tiered early warning instructions and path adjustment suggestions in the execution module are as follows: R1. Receive the predictive risk map and recommended route plan, simultaneously acquire the real-time location coordinates and driving trajectory data of the dangerous goods transport vehicle, extract the real-time dynamic risk value of the current location of the dangerous goods transport vehicle and the grid cells of the adjacent route from the predictive risk map, and calculate the current comprehensive risk exposure value by combining the vehicle speed, axle temperature and tank pressure data collected by the real-time sensors of the dangerous goods transport vehicle. The current comprehensive risk exposure value is derived by weighted fusion of real-time dynamic risk value and sensor data deviation value, and is used to quantify the immediate threat level faced by the vehicle.

[0059] R2. Compare the current comprehensive risk exposure value with the preset multi-level risk thresholds. Based on the comparison results, map the threshold range into which the current comprehensive risk exposure value falls to a specific warning level and generate a preliminary warning instruction framework. At the same time, analyze the deviation between the actual driving trajectory of the dangerous goods transport vehicle and the recommended route plan. If the deviation exceeds the preset safety tolerance, trigger the route anomaly flag. The preset multi-level risk thresholds are dynamically adjusted based on historical accident statistics and the cargo hazard level. The optimal value of the preset safety tolerance is determined based on a comprehensive analysis of vehicle dynamics and path tracking error. First, the average lateral deviation and heading angle deviation of the vehicle relative to the planned path at different speeds are statistically analyzed from historical driving data. The two are then weighted and fused to generate a baseline trajectory deviation. Next, a risk weighting coefficient is assigned based on the risk level of the hazardous materials in the current transportation task (such as highly toxic or explosive materials). The baseline trajectory deviation is multiplied by the risk weighting coefficient to obtain the basic value of the dynamic safety tolerance. At the same time, environmental attenuation factors such as real-time road adhesion coefficient and visibility are introduced for correction. The final value must ensure that the vehicle still has sufficient safety margin to avoid entering high-risk areas under extreme conditions.

[0060] R3. Based on the preliminary early warning instruction framework and path anomaly indicators, the system calls the preset rule base to match the corresponding early warning content, response time and notification scope for each specific early warning level, generating structured hierarchical early warning instructions. Simultaneously, it retrieves the avoidance strategy template corresponding to the current specific early warning level and path anomaly indicators from the preset path adjustment strategy base, and combines real-time traffic data and road constraints to parameterize and instantiate the avoidance strategy template to form path adjustment suggestions.

[0061] The pre-defined rule base is a multi-dimensional decision tree-structured database built upon expert knowledge and historical accident cases. This database uses the warning level as the primary index, linking downwards to warning content templates, response time tiers, and notification scope lists. Warning content templates predefine message formats, key parameters (such as risk location and recommended measures), and push priorities based on risk type (e.g., leaks, fires). The response time tier sets handling time windows from minutes to seconds based on a risk diffusion model. The notification scope list dynamically allocates recipients (e.g., drivers, monitoring centers, emergency departments) according to the risk impact radius. The rule base outputs a structured instruction framework by matching the warning level with the above elements in real time.

[0062] The preset route adjustment strategy library is built by integrating geographical topological constraints, historical emergency cases, and real-time traffic situation data. Its core is a set of avoidance strategy templates. Each template is configured with basic avoidance actions (such as detours, deceleration, and stopping), resource constraints (such as road weight and height restrictions), and execution parameters (such as alternative route radius and safe speed) for specific risk levels and route anomaly indicators (such as deviation and congestion). The strategy library receives real-time traffic data (such as accident points and congestion indices) and road constraints (such as traffic control rules). Through the parameter replacement engine, it instantiates the variables in the template into specific coordinates, speed values, and direction commands. During the instantiation process, the feasibility of the strategy and the real-time vehicle status (such as remaining range and load) is simultaneously verified, and finally, an executable route adjustment suggestion is generated.

[0063] A method for intelligent supervision of remote electronic escort of dangerous goods road transport includes the following steps: T1. Collect multi-source data from dangerous goods transport vehicles in real time through vehicle-mounted IoT terminals and preset cloud interfaces, and preprocess the multi-source data to generate a standardized fusion data stream; T2. Based on the standardized fusion data stream, predict the probability and risk level of accidents on the route segment where the dangerous goods transport vehicle is located within a preset time period, and generate a predictive risk map. T3. Based on the predictive risk map, combined with road constraints and the transportation task requirements of dangerous goods transport vehicles, generate a risk avoidance path set, calculate the comprehensive safety score and expected travel time of each path in the risk avoidance path set, and output the recommended path scheme. T4. Based on the predictive risk map and recommended route plan, generate graded early warning instructions and route adjustment suggestions, and send the graded early warning instructions and route adjustment suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while tracking the actual driving trajectory and risk changes of dangerous goods transport vehicles in real time.

[0064] The above are merely embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A remote electronic escort intelligent supervision system for the road transport of dangerous goods, characterized in that: It includes a data acquisition and fusion module, a transportation risk prediction and assessment module, a route planning module, and an execution module; The data acquisition and fusion module is used to collect multi-source data from dangerous goods transport vehicles in real time through the vehicle-mounted IoT terminal and the preset cloud interface, and to preprocess the multi-source data to generate a standardized fused data stream. The transportation risk prediction and assessment module, based on the standardized fusion data stream, predicts the probability and level of accidents on the route segment where the dangerous goods transport vehicle is located within a preset time period, and generates a predictive risk map. The route planning module, based on the predictive risk map and combined with road constraints and the transportation task requirements of hazardous materials transport vehicles, generates a risk avoidance route set, calculates the comprehensive safety score and expected travel time of each path in the risk avoidance route set, and outputs a recommended route plan. The execution module generates tiered early warning instructions and route adjustment suggestions based on the predictive risk map and the recommended route plan, and sends the tiered early warning instructions and route adjustment suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while simultaneously tracking the actual driving trajectory and risk changes of the hazardous materials transport vehicle in real time.

2. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 1, characterized in that: The multi-source data in the data acquisition and fusion module includes status data of dangerous goods transport vehicles, environmental data, cargo data, and traffic data. The status data includes vehicle speed, axle temperature, and tank pressure; the environmental data includes real-time weather and road conditions; the cargo data includes hazardous material type and leak sensor readings; and the traffic data includes traffic congestion index and accident hotspot information.

3. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 2, characterized in that: The method for generating the predictive risk map in the transportation risk prediction and assessment module is as follows: S1. Based on the original planned route of the hazardous materials transport vehicle, map it to a preset two-dimensional path grid coordinate system to form a reference route. Divide the reference route into multiple continuous path grid units at preset distance intervals. Query a preset historical accident database to obtain historical road segment accident statistics with similar geographical features and traffic flow patterns to the current original planned route. Extract the historical accident frequency and severity data corresponding to each path grid unit. Combined with the inherent risk attributes of hazardous materials, calculate the static risk base of each path grid unit. The static risk base of all path grid units is spatially interpolated in the two-dimensional path grid coordinate system to generate an initial static risk field covering the current and future possible travel areas of the hazardous materials transport vehicle. S2. Based on the initial static risk field, preset risk impact factors are introduced for real-time correction. The risk impact factors include: vehicle operating status factors calculated based on real-time vehicle speed and key component temperature data of dangerous goods transport vehicles; environmental disturbance factors assessed based on real-time meteorological information and road surface adhesion coefficient; and cargo real-time risk factors determined based on the chemical stability of the dangerous goods transported by the dangerous goods transport vehicle and tank sealing sensor readings. A weighted superposition algorithm is used to fuse each of the risk impact factors with the static risk base of its corresponding path grid unit. After fusion calculation, the real-time dynamic risk value of each path grid unit is output to generate a dynamically updated risk field that reflects the current comprehensive risk status. S3. Compare the real-time dynamic risk value of each path grid cell in the dynamically updated risk field with a preset multi-level risk threshold. Based on the comparison result, mark the path grid cells whose real-time dynamic risk value falls into different threshold ranges of the multi-level risk threshold as different risk levels and distinguish them using different preset visual codes. Integrate the risk level markings of all path grid cells to generate a predictive risk map on the two-dimensional path grid coordinate system.

4. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 3, characterized in that: The steps for generating the risk avoidance path set in the path planning module are as follows: Q1. Based on the risk level of each path grid cell in the predictive risk map, identify continuous path segments in the baseline path whose risk level exceeds a preset safety threshold and mark them as high-risk path segments; in the two-dimensional path grid coordinate system, delete the high-risk path segments from the baseline path to form a main path containing a path interruption zone, wherein the path interruption zone is jointly defined by the starting deletion node and the ending deletion node left after the high-risk path segment is deleted; Q2. For each of the path interruption zones, in the two-dimensional path grid coordinate system, with the starting deletion node and the ending deletion node as spatial anchor points, an adjustment path framework is constructed to bypass the high-risk path segment; the adjustment path framework consists of a set of adjustment points pre-generated based on the geographical topological relationship in the vicinity of the benchmark path, and the adjustment points are automatically generated by analyzing the road network connectivity around the path interruption zone to form an adjustment point sequence. Q3. Based on the road constraints of each path within the preset range adjacent to the main path and the transportation requirements of the hazardous materials transport vehicle, optimize and adjust the sequence of adjustment points; dynamically distribute the optimized adjustment points to each road intersection, road segment midpoint, and road end point of the adjustment path framework to form a final adjustment point set; sequentially connect each final adjustment point in the final adjustment point set to generate a local risk avoidance path; seamlessly topologically stitch the local risk avoidance path with the main path to output a complete risk avoidance path, and repeat the above process to generate multiple local risk avoidance paths to form the risk avoidance path set.

5. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 4, characterized in that: The method for constructing the adjustment path framework in step Q2 is as follows: Q201. For each path interruption zone, in the two-dimensional path grid coordinate system, identify all road intersections and connecting road segments within a preset buffer radius around the starting deletion node and the ending deletion node, construct a local road network topology map, and extract key topological features from the local road network topology map, including the connectivity of road intersections, the traffic capacity level of road segments, and road network connectivity indicators, to form a set of road network structure features within the preset buffer radius around the path interruption zone. Q202. Calculate the topological importance score of each road intersection in the local road network topology map. The topological importance score is obtained by weighted summation of the number of connecting roads and the capacity level of the connecting roads at the road intersection. Sum the topological importance scores of all road intersections on the boundary of the path interruption zone and multiply them by a normalization coefficient based on the road network connectivity index to generate a reference value for the number of basic adjustment points. Round the reference value for the number of basic adjustment points up. The rounded reference value for the number of basic adjustment points is not lower than a preset minimum adjustment point threshold. Output the number of basic adjustment points. Q203. Based on the topological importance score of each road intersection in the local road network topology map around the path interruption area, the road intersections are classified and screened; the top N road intersections with the highest topological importance scores are determined as core adjustment points, and the core adjustment points constitute the main nodes of the path detour. If the number of core adjustment points is insufficient, auxiliary adjustment points are selected from the remaining road intersections whose connectivity exceeds a preset connectivity threshold and whose traffic capacity level exceeds a preset level. All core adjustment points and auxiliary adjustment points are sorted according to their spatial topological relationship with the starting deletion node and the ending deletion node to form a final adjustment point sequence. The starting deletion node is used as the path starting point, and each point in the final adjustment point sequence is connected in sequence. The ending deletion node is used as the endpoint to generate an adjustment path framework.

6. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 5, characterized in that: The road constraints mentioned in step Q3 include road grade restrictions, load limits, and traffic control rules; the transportation task requirements for dangerous goods transport vehicles include the expected arrival time window and the hazard level of the goods. The specific content of optimizing and adjusting the adjustment point sequence is as follows: Based on the adjustment point sequence, combined with the road grade restrictions, load restrictions, and traffic control rules, calculate the road traffic feasibility index for each adjustment point; simultaneously, based on the expected arrival time window and cargo hazard level, evaluate the time window matching degree and risk control priority of each adjustment point; input the road traffic feasibility index, the time window matching degree, and the risk control priority into a preset multi-objective optimization function, and generate a comprehensive adjustment coefficient for each adjustment point through weighted summation; based on the level of the comprehensive adjustment coefficient, rearrange the adjustment points in the adjustment point sequence, remove the adjustment points whose comprehensive adjustment coefficient is lower than a preset feasibility threshold, and output the optimized and adjusted adjustment point sequence.

7. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 6, characterized in that: The method for outputting the recommended path scheme in the path planning module is as follows: Q4. For each local risk avoidance path in the risk avoidance path set, extract its path geometric features and risk distribution features. The path geometric features include the total path length, the sum of turning angles, and the average radius of curvature. The risk distribution features are characterized by the average and maximum values ​​of the real-time dynamic risk values ​​of all the path grid cells traversed by the local risk avoidance path. Map the path geometric features and the risk distribution features to time cost coefficients and risk exposure coefficients, respectively. Combined with the expected arrival time window in the transportation task requirements of dangerous goods transport vehicles, calculate the time compliance score of the local risk avoidance path. Calculate the comprehensive safety score and expected travel time for each local risk avoidance path based on the time cost coefficient, risk exposure coefficient, and time compliance score. Q5. Standardize the comprehensive safety score and expected travel time of all local risk avoidance paths in the risk avoidance path set, and use the Pareto optimal ranking algorithm to sort each local risk avoidance path in multiple dimensions. Select the local risk avoidance paths with the top 30% comprehensive safety scores and expected travel times not exceeding a preset time threshold as the recommended path set. The recommended path set is the recommended path scheme.

8. The intelligent monitoring system for remote electronic escort of dangerous goods road transport according to claim 7, characterized in that: The steps for generating the tiered early warning command and the path adjustment suggestion in the execution module are as follows: R1. Receive the predictive risk map and the recommended route plan, simultaneously acquire the real-time location coordinates and driving trajectory data of the dangerous goods transport vehicle, extract the current location of the dangerous goods transport vehicle and the real-time dynamic risk value of the adjacent path grid unit from the predictive risk map, and calculate the current comprehensive risk exposure value by combining the vehicle speed, axle temperature and tank pressure data collected by the real-time sensors of the dangerous goods transport vehicle. R2. Compare the current comprehensive risk exposure value with the preset multi-level risk thresholds. Based on the comparison results, map the threshold range into which the current comprehensive risk exposure value falls to a specific warning level and generate a preliminary warning instruction framework. At the same time, analyze the deviation between the actual driving trajectory of the dangerous goods transport vehicle and the recommended route plan. If the deviation exceeds the preset safety tolerance, trigger the route anomaly flag. R3. Based on the preliminary warning instruction framework and the path anomaly flag, call the preset rule base to match the corresponding warning content, response time and notification scope for each specific warning level, and generate a structured hierarchical warning instruction. Simultaneously, the system retrieves avoidance strategy templates corresponding to the current specific warning level and the path anomaly indicator from the preset path adjustment strategy library. Combining real-time traffic data and road constraints, the avoidance strategy templates are parameterized and instantiated to form path adjustment suggestions.

9. A method for intelligent supervision of remote electronic escort of dangerous goods road transport according to claim 8, characterized in that: Includes the following steps: T1. Collect multi-source data of dangerous goods transport vehicles in real time through vehicle-mounted IoT terminals and preset cloud interfaces, and preprocess the multi-source data to generate a standardized fusion data stream; T2. Based on the standardized fused data stream, predict the probability and risk level of accidents on the route segment where the dangerous goods transport vehicle is located within a preset time period in the future, and generate a predictive risk map. T3. Based on the predictive risk map, combined with road constraints and the transportation task requirements of dangerous goods transport vehicles, generate a risk avoidance path set, calculate the comprehensive safety score and expected travel time of each path in the risk avoidance path set, and output a recommended route plan. T4. Based on the predictive risk map and the recommended route plan, generate graded early warning instructions and route adjustment suggestions, and send the graded early warning instructions and route adjustment suggestions to the vehicle-mounted IoT terminal and the preset monitoring center, while tracking the actual driving trajectory and risk changes of the dangerous goods transport vehicle in real time.

Images