Hazardous chemical cargo transport vehicle supervision method and system

Abstract

The invention discloses a hazardous chemical substance cargo transport vehicle supervision method and system, and relates to the technical field of vehicle supervision, and the method comprises the steps: dividing a hazardous chemical substance region in a transport vehicle, collecting the environment data of the divided hazardous chemical substance region, and predicting the risk behavior of hazardous chemical substances in the transportation process based on the environment data; action behaviors, associated with the risk behaviors of the hazardous chemical substances, of the target group on the cargo transportation path are extracted, risk assessment is carried out on the action behaviors of the target group, risk assessment results of the target group are obtained, and the target group comprises behaviors, motor vehicles and non-motor vehicles. According to the method, when the diffusion prediction model is optimized, the dynamic adaptability of the model can be enhanced by introducing the time factor, it is ensured that the model can accurately predict the diffusion trend of the hazardous chemical substance in different time periods, and therefore a more stable and accurate prediction result is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle supervision, and in particular to a method and system for supervising vehicles transporting hazardous chemicals. Background Art

[0002] Hazardous chemicals are chemicals that may pose significant risks to people, the environment, or property during production, transportation, storage, or use. These include substances with hazardous properties such as flammable, explosive, toxic, corrosive, radioactive, and oxidizing properties. These substances are classified as hazardous chemicals because they could cause significant harm to the surrounding environment and personnel in the event of a leak, explosion, fire, or other accident.

[0003] Since hazardous chemicals are flammable, explosive, toxic, and corrosive, leaks or accidents can not only cause serious environmental pollution, but can also endanger the lives of people around the transport vehicles and public safety. Supervision ensures the safe transportation of hazardous chemicals during transportation, avoiding accidents caused by vehicle failures, improper operation, external environmental influences, and other factors. However, if hazardous chemicals are transported and collide with pedestrians or motor vehicles, or if accidents occur with other traffic participants in densely populated areas, it will be difficult to respond quickly or take appropriate measures in the event of a hazardous chemical leak or other emergency without being able to integrate with the environment along the route. This increases the probability and risk of accidents.

[0004] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0005] In response to the problems in the related art, the present invention proposes a method and system for supervising hazardous chemical cargo transportation vehicles to overcome the above-mentioned technical problems existing in the existing related art.

[0006] To this end, the specific technical solutions adopted in the present invention are as follows:

[0007] According to one aspect of the present invention, a method for supervising hazardous chemical cargo transport vehicles is provided, the method comprising:

[0008] Divide hazardous chemical areas in transport vehicles, collect environmental data of the divided hazardous chemical areas, and predict the risk behavior of hazardous chemicals during transportation based on the environmental data;

[0009] Extract the action behaviors of the target group on the cargo transportation route that are associated with the risk behaviors of hazardous chemicals, and conduct risk assessment on the action behaviors of the target group respectively to obtain the risk assessment results of the target group;

[0010] Formulate transport vehicle control measures based on the risk assessment results of the target group, and optimize the transport vehicle routes in accordance with the control measures.

[0011] Preferably, dividing the hazardous chemicals area in the transport vehicle, collecting environmental data of the divided hazardous chemicals area, and predicting the risk behavior of hazardous chemicals during transportation based on the environmental data include:

[0012] Divide the internal area of ​​the transport vehicle according to the characteristic parameters of hazardous chemicals to obtain the divided hazardous chemical area;

[0013] Use sensor groups to collect environmental data from the designated hazardous chemical area, and determine whether hazardous chemicals are engaging in risky behavior based on the environmental data. If so, proceed to the next step; otherwise, continue monitoring the environmental data.

[0014] Establish a diffusion prediction model based on environmental data, and use the diffusion prediction model to predict the dynamic diffusion behavior of hazardous chemicals during transportation;

[0015] The diffusion prediction model is introduced into the pre-built label tree model, and environmental factors are introduced into the label tree model so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments.

[0016] Preferably, a diffusion prediction model is established based on environmental data, and the dynamic diffusion behavior of hazardous chemicals during transportation is predicted by the diffusion prediction model, including:

[0017] Extract environmental measurement data to establish a diffusion prediction model for hazardous chemicals, output the dynamic distribution of hazardous chemicals through the diffusion prediction model, and form a Gaussian distribution based on the dynamic distribution of hazardous chemicals;

[0018] Determine the number of hazardous chemicals in the divided hazardous chemical area. If the number of hazardous chemicals is greater than or equal to the preset threshold, then randomly superimpose the Gaussian distribution of hazardous chemicals on the combined effect of several hazardous chemical source diffusion points to obtain an optimized diffusion prediction model. Otherwise, the diffusion prediction model is used to predict the dynamic diffusion behavior of hazardous chemicals in the future time period.

[0019] Solve the optimized diffusion prediction model to obtain the diffusion coefficient between adjacent hazardous chemical sources, and use the average value of the diffusion coefficient as the parameter of the optimized diffusion prediction model;

[0020] The time factor is introduced into the parameters of the optimized diffusion prediction model to obtain a time-weighted diffusion prediction model and output a stable diffusion field. The dynamic diffusion behavior of hazardous chemicals in the future time period is analyzed based on the stable diffusion field.

[0021] Preferably, the diffusion prediction model is introduced into the pre-built marker tree model, and environmental factors are introduced into the marker tree model, so that the diffusion prediction model predicts the dynamic diffusion behavior of hazardous chemicals in different environments, including:

[0022] Based on the characteristic parameters of hazardous chemicals as target variables, the target variables are trained using a predefined gradient boosting tree to obtain a trained labeling model tree, and the environmental factors suitable for hazardous chemicals are calculated through the labeling model tree;

[0023] The output results of the diffusion prediction model are extracted as the input of the labeling tree model, and the tree depth, minimum sample splitting, and maximum number of leaf nodes of the labeling tree model are adjusted in sequence to enable the diffusion prediction model to perform dynamic predictions based on changes in environmental factors.

[0024] Preferably, the action behaviors of the target group on the cargo transportation route that are associated with the risk behaviors of hazardous chemicals are extracted, and risk assessments are performed on the action behaviors of the target group respectively, and the risk assessment results of the target group are obtained, including:

[0025] The spatial clustering algorithm is used to cluster the actions of the target groups on the cargo transportation path that are affected by the dynamic diffusion behavior of hazardous chemicals. The target groups include behaviors, motor vehicles, and non-motor vehicles.

[0026] Identify high-risk areas along cargo transportation routes;

[0027] A risk assessment model is established based on the target group's actions and behaviors to assess the risk index when the target group's actions and behaviors come into contact with high-risk areas;

[0028] The risk index is compared with the preset risk threshold, and the risk level of the target group is divided according to the comparison results to obtain the risk assessment results of the target group.

[0029] Preferably, the action behaviors of the target group on the cargo transportation path that are affected by the dynamic diffusion behavior of hazardous chemicals are clustered using a spatial clustering algorithm, including:

[0030] Identify the trajectory points on the cargo transportation route where the target group is affected by the dynamic diffusion behavior of hazardous chemicals, and calculate the number of points contained in the neighborhood of the trajectory points;

[0031] Compare the number of points with the MinPts parameter. If the number of points in the neighborhood of the current trajectory point is greater than the MinPts parameter, it means that the current trajectory point is a core point, and all trajectory points in the neighborhood are connected to the core point to form a clustering result. Otherwise, the current trajectory point is marked as an outlier and removed.

[0032] The activity area of ​​the target group is calculated based on the time difference and distance difference between adjacent trajectory points in the same clustering result, and the target group's movement behavior is analyzed according to the time the target group stays in the same activity area.

[0033] Preferably, a risk assessment model is established based on the target group's action behaviors to assess the risk index when the target group's action behaviors come into contact with high-risk areas, including:

[0034] A single-agent risk assessment model is constructed based on pedestrian behavior, and the risk index of pedestrian contact with hazardous chemicals when pedestrians pass through high-risk areas is analyzed using the single-agent risk assessment model.

[0035] A multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when they pass through high-risk areas is analyzed through the multi-agent risk assessment model.

[0036] Preferably, a multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when the motor vehicles and non-motor vehicles pass through high-risk areas is analyzed by the multi-agent risk assessment model. The risk index includes:

[0037] Taking the actions of motor vehicles and non-motor vehicles as participants, the profit functions of motor vehicles and non-motor vehicles are calculated respectively using non-cooperative game strategies.

[0038] Taking motor vehicles and non-motor vehicles as intelligent agents and setting the agent's strategy set, defining the agent's payoff function to construct a multi-agent risk assessment model. In the multi-agent risk assessment model, Nash equilibrium is used to select the optimal strategy of each agent from the strategy set.

[0039] Reinforcement learning is used to train a multi-agent risk assessment model so that it adopts the optimal strategy in high-risk areas. When the optimal strategy is adopted, the risk index of the agent's contact with hazardous chemicals is calculated separately.

[0040] Preferably, the expression of the diffusion prediction model is:

[0041]

[0042] Where C(x,y,t) represents the functional form of the diffusion prediction model; Q represents the source intensity of the released hazardous chemicals; σ x and σ y They represent the diffusion coefficients in the horizontal direction x and the vertical direction y respectively; (x0, y0) represents the diffusion source coordinates of hazardous chemicals; exp represents the exponential function.

[0043] According to another aspect of the present invention, a hazardous chemicals cargo transport vehicle supervision system is provided, the system comprising:

[0044] The hazardous chemicals risk prediction module is used to divide hazardous chemicals areas in transport vehicles, collect environmental data of the divided hazardous chemicals areas, and predict the risk behavior of hazardous chemicals during transportation based on the environmental data;

[0045] The transport vehicle route analysis module is used to extract the action behaviors of the target group on the cargo transport route that are associated with the risk behaviors of hazardous chemicals, and conduct risk assessments on the action behaviors of the target group to obtain the risk assessment results of the target group;

[0046] The transport vehicle control module is used to formulate transport vehicle control measures based on the risk assessment results of the target group and optimize the transport vehicle routes according to the control measures.

[0047] The beneficial effects of the present invention are:

[0048] 1. The diffusion prediction model established based on environmental data in the present invention can simulate the dynamic diffusion behavior of hazardous chemicals during transportation through real-time collected environmental parameters, thereby better predicting possible leakage or diffusion during transportation, timely warning of changes in dangerous areas, and adjusting emergency plans. In addition, the diffusion prediction model based on Gaussian distribution morphology can predict the diffusion behavior of hazardous chemicals under different environments by simulating the diffusion path and concentration distribution of hazardous chemicals, further improving the accuracy of emergency response.

[0049] 2. When optimizing the diffusion prediction model, the present invention introduces a time factor to enhance the dynamic adaptability of the model, ensuring that the model can accurately predict the diffusion trend of hazardous chemicals in different time periods, thereby providing more stable and accurate prediction results. By combining the diffusion prediction model with the marker tree model, dynamic adjustments are made based on environmental factors to improve the flexibility and response capabilities of the model. Moreover, through training and optimization, the marker tree model can provide personalized risk assessments based on the characteristics of hazardous chemicals and environmental conditions, thereby optimizing transportation route planning and emergency response strategies, and further reducing risks during transportation.

[0050] 3. The present invention uses a spatial clustering algorithm to cluster the movement behaviors of target groups on the cargo transportation path, which can effectively identify the specific behavior patterns of different groups on the transportation path, and further analyze the spatial distribution of these behaviors under the influence of the dynamic diffusion of hazardous chemicals. This not only helps to identify the high-density activity areas of the target groups, but also reveals the dynamic risk exposure of the groups, helping to manage risks at key locations. At the same time, the risk assessment model constructed based on the movement behaviors of the target groups can accurately calculate the risk index of each target group when it comes into contact with high-risk areas, thereby providing specific data support for risk control.

[0051] 4. The present invention uses single-agent and multi-agent risk assessment models to independently analyze the risk of exposure to hazardous chemicals for different participants (such as pedestrians, motor vehicles, and non-motor vehicles) when passing through high-risk areas, ensuring that the risk assessment of each group is targeted and personalized. This precise risk assessment helps identify the potential risks of specific groups at specific times and locations, and take effective control measures in advance to avoid accidents. At the same time, combined with game theory methods such as non-cooperative games and Nash equilibrium, it can optimize the decision-making process of group behavior, allowing different participants to make optimal decisions in high-risk areas based on environmental changes and their own interests, thereby minimizing the risk of contact with hazardous chemicals. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 This is a flow chart of a method for supervising hazardous chemical cargo transportation vehicles according to an embodiment of the present invention;

[0054] Figure 2 The present invention is a block diagram of a hazardous chemicals cargo transport vehicle monitoring system.

[0055] In the picture:

[0056] 1. Hazardous chemicals risk prediction module; 2. Transport vehicle route analysis module; 3. Transport vehicle management and control module. DETAILED DESCRIPTION

[0057] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0058] According to an embodiment of the present invention, a method and system for supervising vehicles transporting hazardous chemicals are provided.

[0059] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an embodiment of the present invention, the method for supervising hazardous chemical cargo transportation vehicles includes:

[0060] S1. Divide the hazardous chemicals area in the transport vehicle, collect environmental data of the divided hazardous chemicals area, and predict the risk behavior of hazardous chemicals during transportation based on the environmental data.

[0061] Among them, the hazardous chemicals areas in the transport vehicles are divided, the environmental data of the divided hazardous chemicals areas are collected, and the risk behaviors of hazardous chemicals in the transportation process are predicted based on the environmental data, including:

[0062] The internal area of ​​the transport vehicle is divided according to the characteristic parameters of the hazardous chemicals to obtain the divided hazardous chemicals area.

[0063] It should be noted that the internal areas of the transport vehicle are divided according to the characteristic parameters of hazardous chemicals. The hazardous chemical areas after division include:

[0064] Step 1: Collect and organize the characteristic parameters of hazardous chemicals:

[0065] The characteristic parameters of hazardous chemicals generally include their physical properties, chemical properties, hazardous characteristics, etc. Collecting and organizing these parameters is crucial for demarcating the interior areas of transport vehicles.

[0066] Physical properties:

[0067] Density: The density of a hazardous chemical determines whether it requires special containers or protective measures.

[0068] Boiling point and flash point: affect the volatility and fire and explosion hazards of hazardous chemicals.

[0069] Smell, color: helpful for sensory monitoring and leak detection.

[0070] Viscosity: affects the flowability of liquids, which in turn affects container design and leak prevention.

[0071] Chemical properties:

[0072] Corrosiveness: Some hazardous chemicals may corrode containers and other substances.

[0073] Reactivity: Hazardous chemicals may react with other substances to produce heat, gases, or explosions.

[0074] Volatility: Determines whether hazardous chemicals will evaporate into the air, increasing the risk of leakage or poisoning.

[0075] Hazardous characteristics:

[0076] Toxicity: The degree of harm to humans or the environment.

[0077] Flammability or Explosiveness: Flammable hazardous chemicals require special isolation areas.

[0078] Radioactive: Requires a dedicated protected area and container.

[0079] Step 2: Divide the area according to characteristic parameters:

[0080] Based on the different characteristics of hazardous chemicals, the interior of the transport vehicle is divided into reasonable areas. The interior of the transport vehicle is divided into different hazard levels. The specific basis for division may include the following factors:

[0081] Chemical reactivity: Separate highly reactive hazardous chemicals from other items to avoid unexpected reactions.

[0082] Fire and explosion hazards: Flammable and explosive hazardous chemicals should be separated from other items and placed in special fire-proof areas.

[0083] Volatility: For highly volatile hazardous chemicals, tightly sealed containers or areas should be set up and appropriate ventilation systems should be installed.

[0084] Toxicity and corrosiveness: Set up special areas for highly toxic and corrosive substances and equip them with appropriate protective facilities.

[0085] Regional division model:

[0086] Safety isolation: Hazardous chemicals are isolated according to their hazardous characteristics to ensure that flammable, explosive, highly corrosive or highly reactive chemicals do not come into contact with each other.

[0087] Ventilation and temperature control: For volatile hazardous chemicals, a good ventilation system needs to be set up; for temperature-sensitive substances (such as cryogenic liquids), a temperature control area needs to be set up.

[0088] Signs and warnings: Each divided area should have clear signs and warnings to indicate the type of hazardous chemicals stored in the area and the corresponding safety operation requirements.

[0089] Among them, the internal area of ​​the transport vehicle is divided according to the characteristic parameters of hazardous chemicals, and the clustering algorithm is used to help effectively divide the safety area according to the characteristics of hazardous chemicals to ensure that no dangerous cross-reaction of substances occurs during transportation. Specifically, it includes:

[0090] Hazardous chemicals are divided into different clusters (areas) based on characteristic parameters (such as hazard, reactivity, density, volatility, etc.).

[0091] The number of clusters needs to be specified in advance, and the center of each cluster is determined by calculating the similarity in the feature space.

[0092] The sensor group is used to collect environmental data of the divided hazardous chemical area, and based on the environmental data of the hazardous chemical area, it is determined whether the hazardous chemicals have risky behavior. If there is risky behavior, the next step is executed. Otherwise, the environmental data is continuously monitored.

[0093] It should be noted that using sensor groups to collect environmental data from the divided hazardous chemical area and judging whether there are risky behaviors of hazardous chemicals based on the environmental data of the hazardous chemical area include:

[0094] Step 1: Install a set of sensors in different areas of the transport vehicle to collect real-time environmental data. The types and functions of these sensors should be selected based on the characteristics of the hazardous chemicals, including:

[0095] Temperature sensor: monitors temperature changes during transportation, especially high temperatures that may cause volatilization or reaction of hazardous chemicals.

[0096] Humidity sensor: For some hazardous chemicals (such as dust or gas), changes in humidity may affect their stability or trigger a reaction.

[0097] Gas sensor: used to monitor the concentration of toxic or flammable gases, such as ammonia, sulfur dioxide, methane, etc.

[0098] Pressure sensor: monitors pressure changes in containers during transportation to avoid leakage due to excessive pressure.

[0099] Volatile organic compound (VOCs) sensors: used to detect signs of gas leaks, especially volatile hazardous chemicals.

[0100] Fire detectors: monitor possible fire risks, especially areas with flammable and explosive materials.

[0101] Step 2: The data collected by the sensors needs to be pre-processed and analyzed in real time to identify potential risk behaviors in a timely manner:

[0102] Leverage machine learning models, rule engines, or threshold detection methods to analyze sensor data in real time and identify unusual changes in the environment, such as:

[0103] If the temperature exceeds the predetermined threshold, it indicates that an abnormality has occurred during transportation, which may lead to dangerous reactions of certain hazardous chemicals.

[0104] If the gas concentration exceeds safe levels, it indicates a possible leak.

[0105] If the pressure is too high or too low, there may be a risk of the container rupturing or leaking.

[0106] Rule-based model: Threshold rules are set based on different hazardous chemical characteristics (such as temperature, gas concentration, pressure, etc.). When sensor data exceeds these thresholds, it is determined to be a risky behavior. For example:

[0107] If the temperature exceeds the set threshold, it is judged that there may be a risk of overheating or chemical reaction.

[0108] When the gas concentration exceeds the set standard, it is judged that there may be a risk of leakage or explosion.

[0109] Machine learning-based models: For example, supervised learning models (such as decision trees and support vector machines) are used to train historical data to establish a more accurate risk prediction model, which inputs environmental data and outputs whether risky behavior exists.

[0110] A diffusion prediction model is established based on environmental data, and the dynamic diffusion behavior of hazardous chemicals during transportation is predicted through the diffusion prediction model.

[0111] Among them, a diffusion prediction model is established based on environmental data. The dynamic diffusion behavior of hazardous chemicals during transportation is predicted by the diffusion prediction model, including:

[0112] Environmental measurement data are extracted to establish a diffusion prediction model for hazardous chemicals. The dynamic distribution of hazardous chemicals is output through the diffusion prediction model, and a Gaussian distribution pattern is formed based on the dynamic distribution of hazardous chemicals.

[0113] The diffusion prediction model is a Gaussian distribution model, which is used to predict the diffusion behavior of hazardous chemicals. The mathematical expression of diffusion can generally be expressed using the Gaussian diffusion equation, which describes the spatial distribution of hazardous chemical gases under given conditions. The expression of the diffusion prediction model is:

[0114]

[0115] Where C(x,y,t) represents the concentration of hazardous chemicals at position (x,y) and time t; Q represents the source intensity of the released hazardous chemicals (i.e., the amount of hazardous chemicals released per unit time); σ x and σ y They represent the horizontal and vertical diffusion coefficients respectively, representing the range of diffusion; (x0, y0) represents the diffusion source coordinates of hazardous chemicals; exp represents the exponential function, that is, the concentration decreases rapidly with increasing distance.

[0116] It should be noted that the Gaussian distribution model is used to obtain the dynamic distribution of hazardous chemicals in space. At this time, the concentration of hazardous chemicals will change with time and space. For example, assuming that the initial hazardous chemical source is located at position (x0, y0) = (0, 0) and diffuses within a certain time interval, for example:

[0117] Q = 100g / s (source intensity, i.e. the amount of hazardous chemicals released per unit time).

[0118] σ x =10m,σ y =10m (diffusion coefficient, determines the range of diffusion).

[0119] At this point, the Gaussian diffusion equation can be used to predict the concentration C(x, y, t) of hazardous chemicals at different locations. For example, at time t = 0, the concentration of hazardous chemicals at a certain location is calculated. If the target location is (x = 5, y = 5), substitute it into the diffusion prediction model;

[0120]

[0121] Then, we can get the concentration value at the point (5,5).

[0122] The number of hazardous chemicals in the divided hazardous chemicals area is determined. If the number of hazardous chemicals is greater than or equal to the preset threshold, the optimized diffusion prediction model is obtained by randomly superimposing the joint effect of several hazardous chemical source diffusion points based on the Gaussian distribution of hazardous chemicals. Otherwise, the dynamic diffusion behavior of hazardous chemicals in the future time period is predicted through the diffusion prediction model.

[0123] It should be noted that if the quantity of hazardous chemicals is greater than or equal to the threshold, the optimized diffusion prediction model is obtained by randomly superimposing the joint effect of several hazardous chemical source diffusion points according to the Gaussian distribution morphology.

[0124] The overall diffusion effect is obtained by superimposing the Gaussian distributions of multiple hazardous chemical sources. Assuming there are n hazardous chemical sources, their diffusion can be expressed as the sum of multiple Gaussian distributions:

[0125]

[0126] Where C total (x, y, t) represents the optimized diffusion prediction model; i represents the i-th hazardous substance; n represents the source of hazardous chemicals.

[0127] Solve the optimized diffusion prediction model to obtain the diffusion coefficient between adjacent hazardous chemical sources, and use the average value of the diffusion coefficient as the parameter of the optimized diffusion prediction model;

[0128] The time factor is introduced into the parameters of the optimized diffusion prediction model to obtain a time-weighted diffusion prediction model and output a stable diffusion field. The dynamic diffusion behavior of hazardous chemicals in the future time period is analyzed based on the stable diffusion field.

[0129] It should be noted that the time factor is introduced into the optimization of the diffusion prediction model parameters to obtain a time-weighted diffusion prediction model and output a stable diffusion field. The dynamic diffusion behavior of hazardous chemicals in the future time period is analyzed based on the stable diffusion field, including:

[0130] Step 1: Introduce the time factor into the diffusion prediction model:

[0131] The diffusion process is significantly affected by time, especially when environmental conditions (such as wind speed, temperature, and humidity) change, causing the diffusion behavior to change over time. Therefore, a time factor should be introduced into the diffusion model to enable the model to reflect changes in hazardous chemical concentrations over time.

[0132] Based on the original Gaussian diffusion model, a time factor is added to describe the dynamic changes of the diffusion process. Assuming that the diffusion source is (x0, y0) and the release intensity of the source is Q, the concentration distribution after time t can be expressed as:

[0133]

[0134] Where f(t) represents the time factor introduced, i.e., the change in diffusion behavior over time. The time factor f(t) may be affected by environmental changes and can usually be designed in the following forms:

[0135] Exponential decay factor: f(t) = exp(-λt), where λ is a decay constant that represents the rate of concentration decay over time during the diffusion process. A larger λ indicates a faster diffusion rate and a faster concentration decay.

[0136] Linear factor: f(t) = 1 + αt, where α is a constant representing the linear increase in concentration over time during the diffusion process.

[0137] Environmental adjustment factors: f(t) can be adjusted based on real-time environmental data (such as wind speed, temperature, and humidity). For example, when the wind speed is high, the diffusion rate may be faster, so f(t) can change with the wind speed.

[0138] Step 2: Calculate the time-weighted diffusion field:

[0139] The time-weighted diffusion field represents the dynamic distribution of hazardous chemical concentrations in space over time. To obtain a stable diffusion field, time can be integrated or averaged to obtain the stable state reached by the system after long-term operation.

[0140] Calculation of the stable diffusion field:

[0141] The stable diffusion field can be solved by the time-weighted diffusion model. In the time period T, the diffusion field at each moment is calculated and the weighted average is performed to obtain the concentration distribution in the stable state. The expression is:

[0142]

[0143] Where C stable(x,y) It represents the stable diffusion field, the average concentration distribution in the time period T, and the stable state over a long period of time is obtained by weighted calculation of the concentration distribution at each moment.

[0144] If the time factor is exponentially decaying, the integration result will focus on the recent time period, reflecting the rapid changes in the diffusion process.

[0145] If the time factor increases linearly, it may indicate that the concentration gradually increases during the diffusion process, reflecting a continuous release from some source.

[0146] Step 3: Analysis of stable diffusion field and prediction of future diffusion behavior:

[0147] After obtaining the stable diffusion field, the dynamic diffusion behavior of hazardous chemicals in the future time period is predicted through the diffusion field. The stable diffusion field provides a concentration distribution map under a long-term stable state, which can be used to analyze possible risks in the future. The stable diffusion field can be analyzed in the following ways:

[0148] Identification of high-concentration areas: Based on the stable diffusion field, areas with higher concentrations of hazardous chemicals are identified, which may represent high-risk areas.

[0149] Diffusion direction and speed: By stabilizing the diffusion field, we can understand the main direction and speed of hazardous chemical diffusion. This helps determine the diffusion range and potential impact areas.

[0150] Dynamic change trend: If the diffusion source changes (such as intensity increase, position change, etc.), the future diffusion behavior can be predicted based on the change trend of the stable diffusion field.

[0151] Prediction of diffusion behavior in future time periods:

[0152] Based on the stable diffusion field and time-weighted diffusion model, the diffusion of hazardous chemicals in the future can be predicted. Assume that the stable diffusion field C at the current moment has been obtained. stable(x,y) , then we can perform time deduction based on this field to obtain the concentration distribution at future moments.

[0153] If the meteorological conditions are expected to change significantly in the future (such as increased wind speed), the concentration at future moments can be updated based on the stable diffusion field using the adjusted time factor f(t).

[0154] Assume that the current time is t=0 and the stable diffusion field is C stable(x,y) , and the wind speed is expected to increase in the next hour, the effect of the wind speed increase can be reflected in the time factor f(t), and the diffusion model can be adjusted to predict the concentration distribution of hazardous chemicals in the next hour. The expression is:

[0155] C forecast (x,y,t+1)=C stable(x,y) f(t+1);

[0156] Where f(t+1) represents the time factor adjusted according to wind speed changes.

[0157] The diffusion prediction model is introduced into the pre-built label tree model, and environmental factors are introduced into the label tree model so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments.

[0158] Among them, the diffusion prediction model is introduced into the pre-built label tree model, and the environmental factors are introduced into the label tree model, so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments, including:

[0159] Based on the characteristic parameters of hazardous chemicals as target variables, the target variables are trained using a predefined gradient boosting tree to obtain a trained labeling model tree, and the environmental factors suitable for hazardous chemicals are calculated through the labeling model tree;

[0160] The output results of the diffusion prediction model are extracted as the input of the labeling tree model, and the tree depth, minimum sample splitting, and maximum number of leaf nodes of the labeling tree model are adjusted in sequence to enable the diffusion prediction model to perform dynamic predictions based on changes in environmental factors.

[0161] It should be noted that the diffusion prediction model is introduced into the pre-built marker tree model, and environmental factors are introduced into the marker tree model so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments in practical applications.

[0162] In this task, the goal is to use characteristic parameters of hazardous chemicals (such as hazard and reactivity) as target variables, train the target variables using gradient boosting trees (GBT), and calculate environmental factors suitable for hazardous chemicals using the labeled model tree obtained through training. Next, the output of the diffusion prediction model is used as input, and by adjusting the model's hyperparameters (tree depth, minimum sample split, maximum number of leaf nodes, etc.), the diffusion prediction model can dynamically predict changes in environmental factors. Specifically, the following are involved:

[0163] Step 1: Train the gradient boosting tree model based on the characteristic parameters of hazardous chemicals:

[0164] Gradient boosted trees (GBT) is an ensemble learning method that improves the model's predictive performance by building a series of decision trees and gradually optimizing each tree. GBT is commonly used for regression and classification tasks.

[0165] 1. Target variable selection:

[0166] Select characteristic parameters of hazardous chemicals as target variables, for example:

[0167] Volatility of hazardous chemicals (such as vapor pressure, temperature sensitivity, etc.).

[0168] The reactivity of hazardous chemicals (e.g. reaction rate, chemical stability, etc.).

[0169] Toxicity of hazardous chemicals (such as LD50 value, or related biological effects, etc.).

[0170] The explosiveness or flammability of hazardous chemicals.

[0171] Select one or more characteristic parameters as target variables based on actual needs. Each characteristic parameter will be used as the output of the GBT model.

[0172] 2. Feature selection:

[0173] Feature data can come from different sensor data, historical records, environmental data (such as wind speed, temperature, humidity, etc.), and other factors in the storage / transportation process of hazardous chemicals. Input features include:

[0174] Environmental parameters: such as temperature, humidity, air pressure, etc.;

[0175] Physical and chemical properties: such as density, solubility, etc.;

[0176] Storage / transportation conditions: such as container type, container sealing, whether there is a possibility of chemical reaction, etc.

[0177] 3. Training the gradient boosting tree model includes:

[0178] Initialize the model: Use the mean of the training data or a simple model for initialization.

[0179] Iterative training: In each iteration, a new tree is trained with the goal of minimizing the error between the current model and the actual value (optimized by gradient descent).

[0180] Update the model: By weighting each new tree, the model's prediction results are gradually optimized.

[0181] During training, loss functions such as mean squared error (MSE) or logarithmic loss (LogLoss) are usually used.

[0182] 4. Output the marked model tree:

[0183] After training, a labeled model tree (GBDT model) is obtained. This model uses characteristic parameters as target variables and predicts the characteristic parameters of hazardous chemicals based on different environmental factors (input features). This model will be used as input for the diffusion prediction model in subsequent steps.

[0184] Step 2: Extract the output of the diffusion prediction model as input features:

[0185] The output of the diffusion prediction model can be used to obtain the concentration distribution of hazardous chemicals in space through a Gaussian distribution model or other diffusion models. The diffusion prediction model generates a dynamic hazardous chemical concentration field based on changes in environmental factors (such as temperature, wind speed, humidity, etc.).

[0186] 1. Diffusion prediction model output:

[0187] Spatial distribution: concentration values ​​of hazardous chemicals at various locations.

[0188] Temporal changes: How the concentration of hazardous chemicals changes at a certain point in the future.

[0189] Risk assessment: Based on the diffusion model, evaluate the potential risks in different areas and predict possible high-risk areas.

[0190] These output results will serve as input to the gradient boosting tree model to affect the prediction of hazardous chemicals characteristic parameters.

[0191] 2. Input features of the model:

[0192] Environmental factors: such as temperature, humidity, wind speed, air pressure, etc.

[0193] Output of the diffusion model: concentration values ​​of hazardous chemicals at different locations and time points.

[0194] Spatial information: the main direction, speed, path, etc. of the spread of hazardous chemicals.

[0195] 3. Adjust the hyperparameters of the labeling tree model:

[0196] In order to enable the diffusion prediction model to make dynamic predictions based on changes in environmental factors, it is necessary to adjust the hyperparameters of the labeling tree model to ensure that it can adjust the prediction results according to actual environmental changes.

[0197] 3.1. Adjust the tree depth (Tree Depth):

[0198] The depth of the tree determines the complexity of the model. Deeper trees can capture more complex relationships between features, but may also lead to overfitting. Shallower trees may not be able to fully learn the patterns in the data. The depth of the tree needs to be selected according to the complexity of the data.

[0199] Larger tree depth: can capture more complex patterns and is suitable for data with strong nonlinear relationships.

[0200] Small tree depth: reduces model complexity, prevents overfitting, and is suitable for linear relationships between features and targets.

[0201] 3.2, Minimum Sample Split (Min Samples Split):

[0202] Minimum sample split is a parameter that controls the minimum number of samples for each node. Before a node splits, at least this number of samples must be present to continue splitting. Adjusting this parameter can prevent the model from learning noisy data during training.

[0203] Increasing the minimum number of sample splits: will reduce the complexity of the tree and prevent overfitting.

[0204] Reducing the minimum number of sample splits: This can improve the model's fitting ability, but may cause overfitting.

[0205] 3.3. Maximum number of leaf nodes (Max Leaf Nodes):

[0206] The maximum number of leaf nodes determines the maximum number of leaf nodes that can be in the tree. When the number of leaf nodes is large, the predictive ability of the tree is enhanced, but it may also lead to overfitting of the model.

[0207] Increasing the number of leaf nodes: This will increase the complexity and accuracy of the model, but may cause overfitting.

[0208] Reducing the number of leaf nodes: helps prevent overfitting, but may also reduce the performance of the model.

[0209] Step 4: Dynamic prediction and model application:

[0210] Once the labeled model tree is obtained through gradient boosting tree training and the hyperparameters (tree depth, minimum sample split, maximum number of leaf nodes, etc.) are adjusted, the model can be applied to dynamic prediction of hazardous chemical diffusion, including:

[0211] The real-time diffusion prediction model output data (such as hazardous chemical concentration distribution and risk assessment) is used as the input features of the labeling tree; based on updated environmental factors (such as wind speed, temperature changes, etc.), the model will dynamically predict changes in hazardous chemical characteristics (such as concentration, reactivity, etc.).

[0212] S2. Extract the action behaviors of the target group on the cargo transportation route that are associated with the risk behaviors of hazardous chemicals, and perform risk assessment on the action behaviors of the target group respectively to obtain the risk assessment results of the target group.

[0213] Among them, the action behaviors of the target group on the cargo transportation route that are associated with the risk behaviors of hazardous chemicals are extracted, and the risk assessment of the action behaviors of the target group is performed respectively, and the risk assessment results of the target group are obtained, including;

[0214] The spatial clustering algorithm is used to cluster the action behaviors of the target groups on the cargo transportation path that are affected by the dynamic diffusion behavior of hazardous chemicals, and the target groups include behaviors, motor vehicles and non-motor vehicles.

[0215] Among them, the spatial clustering algorithm is used to cluster the target group's actions on the cargo transportation path affected by the dynamic diffusion behavior of hazardous chemicals, including:

[0216] Identify the trajectory points on the cargo transportation route where the target group is affected by the dynamic diffusion behavior of hazardous chemicals, and calculate the number of points contained in the neighborhood of the trajectory points;

[0217] Compare the number of points with the MinPts parameter. If the number of points in the neighborhood of the current trajectory point is greater than the MinPts parameter, it means that the current trajectory point is a core point, and all trajectory points in the neighborhood are connected to the core point to form a clustering result. Otherwise, the current trajectory point is marked as an outlier and removed.

[0218] The activity area of ​​the target group is calculated based on the time difference and distance difference between adjacent trajectory points in the same clustering result, and the target group's movement behavior is analyzed according to the time the target group stays in the same activity area.

[0219] It should be noted that the spatial clustering algorithm used is the DBSCAN clustering algorithm. Using the DBSCAN clustering algorithm to cluster the actions of target groups along cargo transportation routes that are affected by the dynamic diffusion of hazardous chemicals effectively divides the target group's actions into different categories and identifies which behaviors are affected by the spread of hazardous chemicals. This is crucial for analyzing the reactions of various groups during hazardous goods transportation, predicting potential safety risks, and implementing effective emergency response measures.

[0220] Identify high-risk areas along cargo transportation routes.

[0221] It should be noted that spatial correlation analysis using GIS software can determine the relationship between hazardous chemical diffusion areas and transportation routes, and identify high-risk areas on cargo transportation routes, including:

[0222] GIS software is used to import hazardous chemical diffusion and transportation route data. Hazardous chemical diffusion data can be modeled using meteorological data, topographic features, and the physical and chemical properties of hazardous chemicals to determine the spatial distribution of hazardous chemical concentrations. Transportation route data includes roads, routes, and traffic information for transport vehicles. Spatial analysis tools can be used to spatially overlay hazardous chemical diffusion areas and transportation routes, identifying which transportation routes cross or approach areas with high hazardous chemical concentrations. Secondly, combining hazardous chemical diffusion patterns with traffic flow, a buffer zone analysis is conducted. This involves setting a buffer zone around the hazardous chemical diffusion area, assessing the overlap of transportation routes within this buffer zone, and identifying high-risk areas.

[0223] A risk assessment model is established based on the target group's action behavior to evaluate the risk index when the target group's action behavior comes into contact with high-risk areas.

[0224] It should be noted that the difference between the risk index of pedestrians and motor vehicles and non-motor vehicles is:

[0225] Pedestrians: The risk index is higher because they are exposed to high-risk areas for a longer time and may be more likely to come into contact with hazardous chemicals due to their slow speed. At the same time, pedestrians have stronger risk avoidance capabilities, but are exposed for longer time when they stop or walk slowly.

[0226] Motor vehicles: The risk index is lower because motor vehicles pass through high-risk areas at a faster speed and the exposure time is shorter; however, the airtightness and poor air circulation of motor vehicles may lead to the risk of exposure of people in the vehicle to hazardous chemicals.

[0227] Non-motor vehicles: The risk index is between that of pedestrians and motor vehicles, with moderate speed and short exposure time, but they have strong risk avoidance capabilities and can easily avoid direct contact with high-risk areas.

[0228] Based on the above differences, the differences between the single-agent model (single-agent risk assessment model) and the multi-agent game model (multi-agent risk assessment model) are:

[0229] Single-agent model: The decision-making process is relatively simple, focusing primarily on its own risk exposure and optimal action. Each agent makes decisions independently, and risk assessment is based on its own behavior and the environment in which it is located.

[0230] Multi-agent game models: The decision-making process is more complex, as each agent must consider not only its own risks and benefits but also the actions of other agents. Game theory comes into play here, and agents may compete, cooperate, or coordinate with each other. For example, consider the situation where multiple pedestrians in the same dangerous area must choose the optimal evacuation route to avoid collisions or congestion.

[0231] Among them, a risk assessment model is established based on the target group's action behavior to evaluate the risk index when the target group's action behavior comes into contact with high-risk areas, including:

[0232] A single-agent risk assessment model is constructed based on pedestrians' motion behaviors, and the risk index of pedestrians' contact with hazardous chemicals when passing through high-risk areas is analyzed through the single-agent risk assessment model.

[0233] It should be noted that a single-agent risk assessment model is constructed based on pedestrians' motion behaviors. When pedestrians pass through high-risk areas, the single-agent risk assessment model is used to analyze the risk index of pedestrians coming into contact with hazardous chemicals, including:

[0234] In a single-agent model, pedestrian behavior usually needs to be modeled through a state space and decision-making mechanism. Based on the pedestrian's movements and reactions, the key elements of the model include:

[0235] State space: describes the current state of the pedestrian, including position, velocity, acceleration, and behavior pattern.

[0236] Action space: defines the possible actions that pedestrians may take (e.g., walk, run, stay, hide, avoid high-density areas, etc.).

[0237] Reward function: Each action is assigned a "reward" that evaluates whether the pedestrian's action reduces or increases risk. For example, taking evasive action in a high-density area will be rewarded more.

[0238] Building a single-agent risk assessment model includes:

[0239] Hazardous Area Identification: Based on the output of the hazardous chemical diffusion model, the scope and concentration of high-risk areas are determined. For example, a Gaussian diffusion model is used to predict the concentration of hazardous chemicals at certain points or areas and identify areas that exceed safety thresholds.

[0240] Risk Index Calculation: The risk index can be calculated by evaluating the probability and concentration of pedestrians' exposure to hazardous chemicals. Assuming pedestrians pass through a certain area, the probability of their exposure to hazardous chemicals and the severity of the exposure are calculated. For example:

[0241] Contact probability: The contact probability is calculated by the degree of overlap between the pedestrian's location and the high-risk area.

[0242] Concentration impact: Based on the concentration of hazardous chemicals, the risk of pedestrian exposure at that concentration is assessed. For example, exposure to high concentration areas may increase the risk of poisoning. The risk index can be calculated using the following formula:

[0243] R i =f(P contact ,C conc )·W area ;

[0244] Where R i represents the risk index of pedestrian i; P contact represents the probability of pedestrians coming into contact with hazardous chemicals; C conc Indicates the concentration of hazardous chemicals in the area; W area Indicates the weight of the area, which can be the danger level of the area or other relevant factors.

[0245] A multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when they pass through high-risk areas is analyzed through the multi-agent risk assessment model.

[0246] Among them, a multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles. The multi-agent risk assessment model is used to analyze the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when they pass through high-risk areas. The risk index includes:

[0247] Taking the actions of motor vehicles and non-motor vehicles as participants, the profit functions of motor vehicles and non-motor vehicles are calculated respectively using non-cooperative game strategies.

[0248] Taking motor vehicles and non-motor vehicles as intelligent agents and setting the agent's strategy set, defining the agent's payoff function to construct a multi-agent risk assessment model. In the multi-agent risk assessment model, Nash equilibrium is used to select the optimal strategy of each agent from the strategy set.

[0249] Reinforcement learning is used to train a multi-agent risk assessment model so that it adopts the optimal strategy in high-risk areas. When the optimal strategy is adopted, the risk index of the agent's contact with hazardous chemicals is calculated separately.

[0250] The risk index is compared with the preset risk threshold, and the risk level of the target group is divided according to the comparison results to obtain the risk assessment results of the target group.

[0251] It should be noted that the multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when they pass through high-risk areas is analyzed through the multi-agent risk assessment model. In practical applications, the following are included:

[0252] It should be noted that transport vehicle control measures are formulated based on the risk assessment results of the target group, and transport vehicle routes are optimized according to the control measures, including:

[0253] Step 1: Establish a non-cooperative game model and calculate the profit functions of motor vehicles and non-motor vehicles:

[0254] In high-risk areas, motor vehicles and non-motor vehicles may compete for the optimal route or adopt different strategies to reduce risks. Non-cooperative games are used to model this competitive relationship.

[0255] 1. Participants (agents):

[0256] Agent 1: Motor vehicle (e.g., car, truck);

[0257] Agent 2: non-motorized vehicles (such as bicycles, motorcycles, and electric vehicles);

[0258] 2. Strategy Space

[0259] Each agent has different strategies to choose from:

[0260] Vehicle strategy set S m :

[0261] Fast Pass: Speed ​​up through high-risk areas;

[0262] Slow down and pass: reduce the speed to pass;

[0263] Detour: Avoid high-risk areas;

[0264] Stop and wait: Wait until the concentration of hazardous chemicals decreases before passing through;

[0265] Non-motor vehicle strategy set S n :

[0266] Fast Pass: Speed ​​up through high-risk areas;

[0267] Slow down and pass: reduce the speed to pass;

[0268] Detour: Avoid high-risk areas;

[0269] Stop and wait: Wait until the concentration of hazardous chemicals decreases before passing through;

[0270] 3. Calculation of the profit function:

[0271] The agent's benefit function U needs to consider the following factors:

[0272] Passing efficiency: faster passing means higher returns, slower or stagnant passing means lower returns.

[0273] Safety: Higher risk of exposure to hazardous chemicals reduces returns.

[0274] Collision risk: Non-motor vehicles and motor vehicles may conflict.

[0275] For example:

[0276] P contact The probability of the agent being exposed to hazardous chemicals in the area

[0277] R collision The probability of an agent colliding with other agents

[0278] T pass is the time it takes for the agent to pass through the area

[0279] α, β, γ are weight factors;

[0280] Then we get:

[0281] U m =U n =α(1-P contact )+β(1-R collision )-γT pass ;

[0282] In the formula, the benefits of different strategies will vary depending on the concentration of hazardous chemicals, traffic conditions, and driving methods.

[0283] Step 2: Build a multi-agent risk assessment model:

[0284] Find the optimal strategy combination in a multi-agent environment so that each agent can pass through the high-risk area with minimal risk, and analyze the optimal strategy selection of the agent by using Nash equilibrium.

[0285] 1. Define the agent's strategy set:

[0286] Strategy set: S = {S m ,S n},in:

[0287] Vehicle strategy set S m ;

[0288] Non-motor vehicle strategy set S n ;

[0289] In the game framework, each agent hopes to minimize its own risk while maximizing traffic efficiency, while other agents will also try to optimize their own strategies.

[0290] 2. Calculate Nash equilibrium:

[0291] U m (S m *,S n *)≥U m (S m ,S n *);

[0292] U n (S m *,S n *)≥U n (S m *,S n );

[0293] In the formula, when all agents choose the optimal strategy S*, no agent can obtain higher benefits by changing its own strategy.

[0294] 3. Solve Nash equilibrium:

[0295] The game matrix is ​​used to represent the agent's strategy choices and benefits, and game theory solutions (such as minimum-maximum optimization or mixed strategy Nash equilibrium) are used to find the optimal strategy combination.

[0296] Step 3: Optimize the strategy using reinforcement learning:

[0297] Due to the complex environment, Multi-Agent Reinforcement Learning (MARL) is used to optimize the agent's strategy selection and make it adaptively make the optimal decision. Specifically, it includes:

[0298] Input: Current environmental conditions (hazardous chemical concentration, traffic flow, driving speed).

[0299] Action: Different strategy options (pass quickly, go slowly, detour, wait).

[0300] Reward: R=U m +U n Among them, less exposure to hazardous chemical areas and less conflict with other agents will result in higher rewards.

[0301] Step 4: Calculate the risk index of the agent:

[0302] The trained model can be used to evaluate the risk index of the agent, which is expressed as:

[0303] R index =P contact ×C conc ;

[0304] Where R index Represents the risk index of the agent; P contact represents the interaction probability between the agent and the hazardous chemicals area, C conc Represents the concentration of hazardous chemicals, where the risk index is different for different strategies:

[0305] Fast pass: R index It may be low, but at high speeds, it may increase the risk of collision.

[0306] Detour: R index It may be close to 0, but the driving efficiency is low.

[0307] Slow down: R index It may be higher, but the risk aversion is enhanced.

[0308] Step 5: Risk level classification:

[0309] Compare the calculated risk index with the preset risk threshold to classify the target group's risk level:

[0310] Low risk (R index <0.2): The agent is basically safe.

[0311] Medium risk (0.2≤R index <0.5): May be briefly exposed to hazardous chemicals, so be vigilant.

[0312] High risk (R index≥0.5): There is a high probability of exposure to hazardous chemicals and emergency avoidance measures must be taken.

[0313] S3. Develop transport vehicle control measures based on the risk assessment results of the target group, and optimize the transport vehicle routes in accordance with the control measures.

[0314] It should be noted that corresponding control measures will be formulated based on the assessment results of different risk levels, with a focus on the control of high-risk areas and time periods:

[0315] Control of high-risk areas:

[0316] Speed ​​limit management: Implement speed limits in high-risk areas to reduce the speed of transport vehicles. Lower speeds increase transport vehicles' hazard avoidance reaction time, reducing the risk of exposure to hazardous chemicals.

[0317] Access restrictions: In certain high-risk areas, it may be necessary to restrict the entry of certain types of transport vehicles (such as those transporting flammable and explosive hazardous materials). For example, to determine the high risk level in certain areas, you can set restricted zones or time periods.

[0318] Road closures and detours: If hazardous chemical concentrations are too high in high-risk areas, roads may need to be closed or detours planned for transport vehicles. Leverage traffic management systems to notify relevant vehicles in advance and prevent them from entering high-risk areas.

[0319] High-risk period control:

[0320] Traffic diversion: Strengthen traffic control and manage traffic flow during high-risk periods (e.g., high wind speeds, inclement weather, etc.). For example, traffic light adjustments and road flow control measures can be used to reduce congestion during high-risk periods and prevent transport vehicles from being stuck in areas with high concentrations of hazardous chemicals.

[0321] Dynamically adjust transportation schedules: Avoid high-risk periods through advance planning. Dynamically adjust transportation plans based on real-time weather, traffic, hazardous chemical dispersion, and other factors. For example, avoid periods of high temperature or wind speed to reduce the spread and risk of hazardous chemicals.

[0322] Monitoring and guidance of transport vehicles:

[0323] Real-time monitoring: Real-time monitoring of transport vehicles. Utilizing sensors, GPS, drones and other technologies, the vehicle's location, driving status, and environmental data (such as hazardous chemical concentrations and meteorological data) are collected in real time to adjust the vehicle's driving route in a timely manner.

[0324] Route optimization guidance: Through real-time traffic information and risk assessment data, the best route guidance is provided to transport vehicles to avoid high-risk areas, reduce parking waiting time, and reduce the possibility of contact with hazardous chemicals.

[0325] According to another embodiment of the present invention, Figure 2 As shown, a hazardous chemical cargo transport vehicle supervision system is also provided, which includes a hazardous chemical risk prediction module 1, a transport vehicle path analysis module 2 and a transport vehicle management and control module 3;

[0326] The hazardous chemicals risk prediction module 1 is connected to the transport vehicle path analysis module 2, and the transport vehicle path analysis module 2 is connected to the transport vehicle control module 3;

[0327] Hazardous chemicals risk prediction module 1 is used to divide hazardous chemicals areas in transport vehicles, collect environmental data of the divided hazardous chemicals areas, and predict the risk behavior of hazardous chemicals during transportation based on the environmental data;

[0328] Transport vehicle route analysis module 2 is used to extract the action behaviors of the target group on the cargo transport route that are associated with the risk behaviors of hazardous chemicals, and to perform risk assessment on the action behaviors of the target group to obtain the risk assessment results of the target group;

[0329] The transport vehicle control module 3 is used to formulate transport vehicle control measures based on the risk assessment results of the target group and optimize the transport vehicle routes according to the control measures.

[0330] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for supervising vehicles transporting hazardous chemicals, characterized in that: The method includes: Divide hazardous chemicals into zones within transport vehicles, collect environmental data from these zones, and predict risk behaviors of hazardous chemicals during transportation based on this environmental data. Extracting the actions and behaviors of target groups on the cargo transportation route that are associated with risk behaviors of hazardous chemicals, and performing risk assessments on the actions and behaviors of the target groups respectively to obtain risk assessment results of the target groups, where the target groups include behaviors, motor vehicles, and non-motor vehicles; Formulate transport vehicle control measures based on the risk assessment results of the target group and optimize transport vehicle routes in accordance with the control measures; Among them, the risk behaviors of hazardous chemicals during transportation predicted based on environmental data include: Establish a diffusion prediction model based on environmental data, and use the diffusion prediction model to predict the dynamic diffusion behavior of hazardous chemicals during transportation; The diffusion prediction model is introduced into the pre-built label tree model, and the environmental factors are introduced into the label tree model so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments; Conduct risk assessments on the target group's actions and behaviors, and obtain the following risk assessment results for the target group: A risk assessment model is established based on the target group's action behavior to evaluate the risk index when the target group's action behavior comes into contact with high-risk areas.

2. A method for supervising hazardous chemicals transport vehicles according to claim 1, characterized in that: The division of hazardous chemical areas in transport vehicles and the collection of environmental data of the divided hazardous chemical areas include: Divide the internal area of ​​the transport vehicle according to the characteristic parameters of hazardous chemicals to obtain the divided hazardous chemicals area; The sensor group is used to collect environmental data of the divided hazardous chemical area, and based on the environmental data of the hazardous chemical area, it is determined whether the hazardous chemicals have risky behavior. If there is risky behavior, the next step is executed. Otherwise, the environmental data is continuously monitored.

3. A method for supervising hazardous chemicals transport vehicles according to claim 1, characterized in that: The diffusion prediction model is established based on environmental data, and the dynamic diffusion behavior of hazardous chemicals during transportation is predicted by the diffusion prediction model, including: Extract environmental measurement data to establish a diffusion prediction model for hazardous chemicals, output the dynamic distribution of hazardous chemicals through the diffusion prediction model, and form a Gaussian distribution based on the dynamic distribution of hazardous chemicals; Determine the number of hazardous chemicals in the divided hazardous chemical area. If the number of hazardous chemicals is greater than or equal to the preset threshold, then randomly superimpose the Gaussian distribution of hazardous chemicals on the combined effect of several hazardous chemical source diffusion points to obtain an optimized diffusion prediction model. Otherwise, the diffusion prediction model is used to predict the dynamic diffusion behavior of hazardous chemicals in the future time period. Solve the optimized diffusion prediction model to obtain the diffusion coefficient between adjacent hazardous chemical sources, and use the average value of the diffusion coefficient as the parameter of the optimized diffusion prediction model; The time factor is introduced into the parameters of the optimized diffusion prediction model to obtain a time-weighted diffusion prediction model and output a stable diffusion field. The dynamic diffusion behavior of hazardous chemicals in the future time period is analyzed based on the stable diffusion field.

4. A method for supervising hazardous chemicals transport vehicles according to claim 1, characterized in that: The method of introducing the diffusion prediction model into the pre-built label tree model and introducing environmental factors into the label tree model so that the diffusion prediction model can predict the dynamic diffusion behavior of hazardous chemicals in different environments includes: Based on the characteristic parameters of hazardous chemicals as target variables, the target variables are trained using a predefined gradient boosting tree to obtain a trained labeling model tree, and the environmental factors suitable for hazardous chemicals are calculated through the labeling model tree; The output results of the diffusion prediction model are extracted as the input of the labeling tree model, and the tree depth, minimum sample splitting, and maximum number of leaf nodes of the labeling tree model are adjusted in sequence to enable the diffusion prediction model to perform dynamic predictions based on changes in environmental factors.

5. A method for supervising hazardous chemicals transport vehicles according to claim 1, characterized in that: The extracting of the action behaviors of the target group on the cargo transportation route that are associated with the risk behaviors of hazardous chemicals, and performing risk assessment on the action behaviors of the target group respectively, to obtain the risk assessment results of the target group includes: The spatial clustering algorithm is used to cluster the action behaviors of the target group on the cargo transportation path affected by the dynamic diffusion behavior of hazardous chemicals; Identify high-risk areas along cargo transportation routes; After establishing a risk assessment model based on the target group's actions and behaviors to assess the risk index when the target group's actions and behaviors come into contact with high-risk areas, the method further includes: The risk index is compared with the preset risk threshold, and the risk level of the target group is divided according to the comparison results to obtain the risk assessment results of the target group.

6. A method for supervising hazardous chemicals transport vehicles according to claim 5, characterized in that: The action behaviors of the target group on the cargo transportation path clustered by the spatial clustering algorithm and affected by the dynamic diffusion behavior of hazardous chemicals include: Identify the trajectory points on the cargo transportation route where the target group is affected by the dynamic diffusion behavior of hazardous chemicals, and calculate the number of points contained in the neighborhood of the trajectory points; Compare the number of points with the MinPts parameter. If the number of points in the neighborhood of the current trajectory point is greater than the MinPts parameter, it means that the current trajectory point is a core point, and all trajectory points in the neighborhood are connected to the core point to form a clustering result. Otherwise, the current trajectory point is marked as an outlier and removed. The activity area of ​​the target group is calculated based on the time difference and distance difference between adjacent trajectory points in the same clustering result, and the target group's movement behavior is analyzed according to the time the target group stays in the same activity area.

7. A method for supervising hazardous chemicals transport vehicles according to claim 6, characterized in that: The risk assessment model established based on the target group's action behavior to assess the risk index when the target group's action behavior contacts a high-risk area includes: A single-agent risk assessment model is constructed based on pedestrian behavior, and the risk index of pedestrian contact with hazardous chemicals when pedestrians pass through high-risk areas is analyzed using the single-agent risk assessment model. A multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when they pass through high-risk areas is analyzed through the multi-agent risk assessment model.

8. A method for supervising hazardous chemicals transport vehicles according to claim 7, characterized in that: The multi-agent risk assessment model is constructed based on the action behaviors of motor vehicles and non-motor vehicles, and the risk index of motor vehicles and non-motor vehicles coming into contact with hazardous chemicals when the motor vehicles and non-motor vehicles pass through high-risk areas is analyzed by the multi-agent risk assessment model. Taking the actions of motor vehicles and non-motor vehicles as participants, the profit functions of motor vehicles and non-motor vehicles are calculated respectively using non-cooperative game strategies. Taking motor vehicles and non-motor vehicles as intelligent agents and setting the agent's strategy set, defining the agent's payoff function to construct a multi-agent risk assessment model. In the multi-agent risk assessment model, Nash equilibrium is used to select the optimal strategy of each agent from the strategy set. Reinforcement learning is used to train a multi-agent risk assessment model so that it adopts the optimal strategy in high-risk areas. When the optimal strategy is adopted, the risk index of the agent's contact with hazardous chemicals is calculated separately.

9. A method for supervising hazardous chemicals transport vehicles according to claim 4, characterized in that: The expression of the diffusion prediction model is: Where C(x,y,t) represents the functional form of the diffusion prediction model; Q represents the source intensity of the released hazardous chemicals; σ x and σ y They represent the diffusion coefficients in the horizontal direction x and the vertical direction y respectively; (x0, y0) represents the diffusion source coordinates of hazardous chemicals; exp represents the exponential function.

10. A hazardous chemical cargo transport vehicle supervision system, used to implement the hazardous chemical cargo transport vehicle supervision method according to any one of claims 1 to 9, characterized in that: The system includes: The hazardous chemicals risk prediction module is used to divide hazardous chemicals areas in transport vehicles, collect environmental data of the divided hazardous chemicals areas, and predict the risk behavior of hazardous chemicals during transportation based on the environmental data; The transport vehicle route analysis module is used to extract the action behaviors of the target group on the cargo transport route that are associated with the risk behaviors of hazardous chemicals, and conduct risk assessments on the action behaviors of the target group to obtain the risk assessment results of the target group; The transport vehicle control module is used to formulate transport vehicle control measures based on the risk assessment results of the target group and optimize the transport vehicle routes according to the control measures.

Images