A method for intelligent access control and risk warning of hazardous chemical vehicles

Abstract

This invention discloses an intelligent entry control and risk warning method for hazardous chemical vehicles, belonging to the field of hazardous chemical transportation safety management technology. The method includes: acquiring multi-source data of vehicles to be entered, including carrier qualification data, vehicle equipment status data, cargo risk data, driver status data, and real-time environmental data; normalizing the data of each dimension to calculate an initial risk score; adaptively adjusting the weight allocation of each level based on dynamic environmental factors; calculating the overall risk value through a weighted comprehensive risk scoring function; comparing the overall risk value with multi-level warning thresholds to generate corresponding control instructions; and generating a comprehensive risk impact factor and forming a closed-loop linkage with on-site scheduling. This invention solves the technical problems of incomplete, static, and disconnected entry risk assessment from subsequent operation and management in existing technologies, achieving comprehensive quantitative assessment, adaptive dynamic adjustment, and closed-loop control of hazardous chemical vehicle entry risks throughout the entire process.

Description

Technical Field

[0001] This invention relates to the technical field of hazardous chemical transportation safety management, specifically to a method for intelligent entry control and risk warning of hazardous chemical vehicles. Background Technology

[0002] Vehicles transporting hazardous chemicals are subject to strict safety inspections and traffic control when entering closed management areas such as chemical industrial parks and parking lots. Currently, with the rapid development of the chemical industry, the number of dedicated hazardous chemical transport vehicles has increased significantly. How to achieve efficient and scientific entry safety assessment and supervision has become a critical issue that urgently needs to be addressed within the industry. The national standard "Construction Specification for Parking Lots for Hazardous Chemical Transport Vehicles in Chemical Industrial Parks" (GB / T45236-2025), implemented on May 1, 2025, clearly stipulates the overall requirements, site selection, overall layout, utilities, safety, fire protection, environmental protection, and information technology construction requirements for parking lots for hazardous chemical transport vehicles in chemical industrial parks. The implementation of this standard further promotes the informatization and intelligent transformation of hazardous chemical vehicle entry control within the industry.

[0003] In recent years, in order to ensure the safety of the transportation and parking of hazardous chemical vehicles, the industry and academia have developed a series of related technical solutions: For example, Chinese patent application CN117953716A discloses a method for safety management of hazardous chemical transport vehicles in chemical industrial parks. This method combines hard measures such as a vehicle management app, internal and external traffic control within the park, and the establishment of dedicated checkpoints with soft management to manage the safety of hazardous chemical transport vehicles. Based on the vehicle management app, real-time location tracking of hazardous chemical transport vehicles is achieved via mobile phone GPS, reducing the purchase and maintenance costs of equipment such as positioning cards and terminals. While this solution focuses on monitoring vehicle trajectories through GPS positioning, it still has shortcomings in the quantitative risk assessment at the entry point.

[0004] Chinese patent application CN117953716A proposes a control method that combines vehicle entry / exit reservation with dedicated checkpoints. This method plans dedicated lanes and traffic signs by acquiring information such as the layout of hazardous chemical enterprises in the chemical industrial park, road grades, and traffic flow. Its core lies in route planning and traffic management, but it does not yet involve a systematic quantitative assessment of the overall risk to vehicles entering the park.

[0005] Chinese patent CN119516752A discloses a safety monitoring and alarm method for highway hazardous materials transport vehicles. This method determines whether hazardous materials sensor data matches preset risk assessment standards. If there is a mismatch, it combines vehicle sensor data and hazardous materials sensor data to predict changes and determine the existence of safety risks, then outputs an alarm. This method focuses on real-time monitoring and trend prediction of hazardous materials sensor data during transportation, but it does not address the comprehensive fusion and evaluation of multi-source heterogeneous data in the entry scenario.

[0006] Chinese patent application CN121686756A discloses a method and system for intelligent monitoring of hazardous chemical vehicles in service areas based on edge computing. This method uses edge computing nodes to acquire real-time location data and status information of hazardous chemical vehicles within the service area to construct dynamic behavioral profiles. It compares the vehicle's trajectory with the planned route on the electronic waybill to determine path consistency and detect clustering anomalies. After generating potential risk signals, it triggers an unmanned aerial vehicle (UAV) system for on-site verification. This solution focuses on the dynamic behavior monitoring and route deviation warnings of vehicles within the service area, falling under the category of post-entry monitoring and not addressing pre-entry risk assessment and access decisions.

[0007] Chinese patent application CN120218616A discloses a knowledge graph-based system and method for monitoring and early warning safety risks in the transportation of hazardous chemical tankers. The system includes a hazardous chemical tanker transportation data acquisition module, a legal and regulatory data and historical accident data acquisition module, a knowledge graph module, and a safety risk situation analysis module, aiming to achieve safety risk analysis and accident cause discovery in the transportation of hazardous chemical tankers. This method focuses on risk analysis of the entire transportation process, emphasizing historical accident data mining and knowledge graph construction, but it is not suitable for real-time quantitative assessment and access control at the entry point.

[0008] Chinese patent application CN119821401A discloses a big data-based method for the safety management of chemical hazardous material vehicles. This method acquires vehicle speed and engine power data, corrects a preset temperature threshold based on average stability, and uses a simulated annealing algorithm to find the optimal solution for the highest vehicle speed during the control period to manage vehicle speed. This method focuses on controlling vehicle speed and differs fundamentally from the technical field of entry risk assessment.

[0009] In addition, the publicly available risk assessment method for hazardous chemical vehicles on highways based on fuzzy set theory uses expert scoring to establish a membership risk assessment matrix of assessment factors and constructs an evaluation fuzzy set based on weighted fuzzy sets. This method introduces fuzzy mathematics for risk assessment, but the determination of weights depends on expert experience, is highly subjective, and is difficult to adapt to dynamically changing entry environments.

[0010] In summary, while existing technical solutions have made progress in some aspects of the safety supervision of hazardous chemical vehicles, they still have the following technical shortcomings: 1. The risk assessment for entry is based on a single dimension and lacks sufficient data integration: Existing technologies often make judgments based on a single dimension (such as vehicle driving status or a single item in hazardous materials sensor data), or use simple rule combinations for risk assessment. They fail to build a quantitative fusion assessment model that covers multiple factors such as vehicle qualifications, personnel status, cargo characteristics, and real-time environment, resulting in limited comprehensiveness and accuracy of risk assessment.

[0011] 2. The risk assessment uses static rules and lacks adaptive adjustment capabilities: Existing technologies mostly employ preset static risk assessment criteria (such as fixed threshold alarms) for risk determination. For example, the method in CN119516752A relies on preset "risk assessment criteria" for matching and judgment, and the assessment threshold lacks a mechanism for dynamic adjustment based on real-time scenarios (such as weather changes and accident incidence patterns at different times). However, the risks of hazardous chemical transportation are closely related to its dynamic environment, and static assessment systems cannot reflect the dynamic changes in the actual risk level, easily leading to false alarms in low-risk scenarios or missed alarms in high-risk scenarios.

[0012] 3. The results of the on-site assessment are disconnected from the on-site operation and management, resulting in a prominent problem of information silos: Existing technologies separate entry inspection from on-site management—once a vehicle completes entry verification, it is removed from the continuous monitoring of the risk assessment system. The assessment information generated at the entry stage fails to play a role in subsequent stages such as parking space allocation, on-site inspection scheduling, and emergency plan matching, creating a "lack of management upon entry" gap in prevention and control. Previous patent applications have lacked a systematic solution to this problem.

[0013] 4. Insufficient quantification of human factors (driver condition) in risk assessment: As one of the core factors in the safety of hazardous chemical transportation, the driver's physiological state, historical violation records, psychological qualities and other "soft indicators" are usually only recorded as qualitative information in the existing technology. There is a lack of algorithmic support to quantify these factors into risk indicators and integrate them with vehicle and cargo factors for evaluation.

[0014] Therefore, there is an urgent need for a method for intelligent entry control and risk warning of hazardous chemical vehicles that can integrate multi-source heterogeneous data, has adaptive dynamic assessment capabilities, and realize the linkage between entry assessment and on-site operation. Summary of the Invention

[0015] This invention provides a method for intelligent entry control and risk warning of hazardous chemical vehicles. It aims to integrate multi-dimensional data such as enterprise qualifications, vehicle equipment, cargo, driver status, and dynamic environment, and achieve adaptive risk assessment through a dynamic weight allocation mechanism. The assessment results are linked with on-site scheduling management in a closed loop to solve the technical problems of incomplete, static, and disconnected entry risk assessment from subsequent operation and management in existing technologies.

[0016] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent entry control and risk warning of hazardous chemical vehicles, comprising the following steps: S1: Steps for acquiring multi-source data: Acquire multi-source data on vehicles carrying hazardous chemicals awaiting entry. This multi-source data includes, but is not limited to, the following five types of data sources: (1) Carrier qualification data: including the carrier’s licensed transportation qualification information, historical safety records, number of safety violations and penalty records in the past three years, enterprise safety management system rating, etc., which are derived from the enterprise management system and industry regulatory database; (2) Vehicle equipment status data: including basic vehicle information (license plate number, vehicle model, vehicle identification number), vehicle annual inspection validity information, tank validity period and pressure vessel inspection certificate information, as well as real-time acquired vehicle sensor data, including but not limited to tire pressure monitoring data, braking system status data, vehicle static grounding status data, etc., which are derived from vehicle sensors and vehicle management system. (3) Risk data of the cargo carried: including electronic waybill information carried with the vehicle in accordance with the State Council's "Regulations on the Safety Management of Hazardous Chemicals" and other regulations, safety data of the hazardous chemicals carried (hazardous characteristics, handling and storage requirements, transportation conditions, etc.), quantity of cargo and packaging method, which are derived from the electronic waybill system and hazardous chemical database; (4) Driver status data: including driver qualification information, driver's license validity period, historical traffic violation records and demerit points, historical accident records, and driver physiological status data (fatigue level, degree of inattention, etc.) obtained in real time through the vehicle behavior detection system, which comes from the driver management system and vehicle behavior monitoring equipment. (5) Real-time environmental data: including real-time weather warning information, road condition data, real-time traffic congestion index, and statistical data on high-incidence periods of accidents in the region, which are released in accordance with the relevant regulations of the meteorological department of the State Council. These data are derived from meteorological API interfaces, traffic monitoring systems, and park management systems.

[0017] S2: Multi-source data standardization and normalization processing steps: The five types of data obtained in S1 are cleaned and normalized, and uniformly mapped to the [0,1] interval. The specific processing method is as follows: For continuous parameters with known membership functions (such as driver fatigue detection scores), a positive scoring function S(y) is established according to industry standards, and the initial basic risk score is calculated using the Min-Max normalization method. For discrete parameters (such as the number of corporate violations and violation points), a discrete level deduction table is established based on authoritative industry grading standards, mapped according to the quantification rules specified in the deduction table, and normalized to the [0,1] interval. For safety or risk level parameters with only qualitative descriptions, they are converted into quantifiable numerical sequences using semantic transformation algorithms before normalization.

[0018] As an example, the normalization mapping for various types of data is based on the following: (a) For driver fatigue levels, the original indicator is the PERCLOS value (the proportion of time the eyes are closed per unit of time). According to the transportation industry standard JT / T1247-2019, the fatigue risk score S_fatigue=min(1,max(0,(PERCLOS-0.15) / (0.45-0.15))), where 0.15 is the upper limit threshold of normal wakefulness and 0.45 is the threshold of severe fatigue. This mapping is linearly normalized and has a clear physiological basis.

[0019] (b) For the number of safety violations by enterprises, in accordance with the scoring rules of the "Safety Production Management System for Road Transport Enterprises of Dangerous Goods", the number of violations in the past three years of 0, 1-2, 3-5, and more than 6 are mapped to 0, 0.25, 0.5, and 1.0 respectively, and then normalized by dividing by the highest score.

[0020] (c) For the hazard class of goods, the hazard classification in GB12268-2012 "List of Dangerous Goods" is used as a reference. High-risk categories such as Class 1 (explosives), Class 2.1 (flammable gases), Class 2.3 (toxic gases), and Class 3 (flammable liquids, flash point <23℃) are mapped to 0.8~1.0, medium-risk categories are mapped to 0.4~0.8, and low-risk categories are mapped to 0~0.4. The specific score is determined by the comprehensive hazard characteristic score.

[0021] The initial basic risk scores for the five types of data sources were ultimately obtained. These correspond to the risk scores for the carrier's qualifications, vehicle and equipment status, cargo, driver status, and real-time environment, respectively.

[0022] S3: Adaptive weight allocation steps based on dynamic environmental factors: Based on a preset weight heuristic matrix M, the weights of each level are adaptively adjusted in combination with real-time dynamic environmental information.

[0023] The preset weight heuristic matrix M is expressed as:

[0024] in, The initial base weights to be adapted are used to represent The basic contribution of each risk score to the overall risk score under normal circumstances can be determined based on the correlation statistics of historical accident data or the experience of industry experts.

[0025] The adaptive weight adjustment process is implemented through the weight adjustment function K(C,t,α), which introduces core dynamic parameters in real time during the weight adjustment process: the real-time environmental correction coefficient C, the time-period risk correction coefficient t, and the cargo risk correction coefficient α, to dynamically adjust the initial base weight w_i_0. After dynamically adjusting the base weights, new dynamic weights are generated. This dynamic weight set is dynamically updated as the input parameters change in real time.

[0026] Define the dynamic weight adjustment function as follows:

[0027] Where i = 1 to 5 correspond to the five dimensions respectively; j represents the type of dynamic factor (such as weather, time period, congestion, cargo hazard level, etc.); Δ_ij is the normalized influence value of the j-th dynamic factor, with a value range of [0,1]; α_ij is the corresponding sensitivity coefficient, which is determined by logistic regression analysis of historical accident data.

[0028] For example, for the real-time environment dimension (i=5): When visibility is <100m, rainfall >10mm / h, or road surface slippage coefficient >0.6, Δ_weather=1, sensitivity coefficient α_5, and weather is taken as 0.8. .

[0029] When the road congestion index (provided by Gaode / Baidu Maps API) > 6.0, Δ_congestion = 1, α_5, and congestion is set to 0.5. Multiply by an additional 1.5.

[0030] For the driver state dimension (i=4): When the current time period falls within the historically high accident incidence period (such as from 23:00 to 5:00 the next day), Δ_time=1, α_4,time is taken as 1.2 (because the accident rate during this period is 3.5 times that during the day, the weight is increased to 2.2 times).

[0031] For the cargo risk dimension (i=3): When the cargo belongs to Class 2.1 (flammable gas), Class 2.3 (toxic gas), or Class 3 liquid with a flash point <23℃ as defined in GB12268, Δ_cargo=1, α_3,cargo is taken as 1.0, then .

[0032] Preset environmental thresholds: Severe weather conditions are defined as follows: visibility <200m or hourly rainfall >10mm or road surface slip coefficient (based on friction coefficient) <0.5.

[0033] Congestion threshold: Real-time traffic congestion index (0~10) > 6.0.

[0034] List of high-risk substances: See the GB12268 classification above and the enterprise's custom high-risk list.

[0035] The specific rules for weight adjustment include, but are not limited to, one or more of the following: When real-time weather data indicates severe weather (such as heavy rain, fog, snow, etc.) or abnormal road conditions, the weight value of S5 is dynamically increased. When the current time period falls within a high-incidence period for accidents identified through statistical analysis (such as 11:00 PM to 5:00 AM the next day, holidays, etc.), dynamic adjustments are made. The weight value; When the road congestion index exceeds the preset congestion threshold, the system will dynamically adjust its settings. The weight value is linked with the vehicle navigation system to send congestion warnings; When the hazard characteristics of goods are identified as highly toxic substances, highly toxic gases, or flammable and explosive high-risk substances, dynamic upgrading is required. The weight value.

[0036] S4: Comprehensive Risk Scoring Steps: Based on the dynamic weights generated after the above weight adjustments... and the initial basic risk score obtained in step S2 The overall risk value of the vehicles to be entered is calculated by using a weighted comprehensive risk scoring function, and the overall risk of the vehicles is quantitatively assessed.

[0037] The weighted comprehensive risk scoring function F is set to a linear form of weighted summation:

[0038] The overall risk value F is ultimately set in the range of [0,1]. The closer the F value is to 1, the higher the overall risk level of the vehicles waiting to enter the site.

[0039] During the evaluation process, the real-time changes of the selected core parameter inputs are continuously monitored, and the F-value is automatically recalculated in a timely manner to maintain the real-time validity of the evaluation results.

[0040] S5: Steps for Multi-Level Early Warning Threshold Matching and Control Instruction Generation: The overall risk value F calculated in step S4 is compared with the pre-set threshold. Compare the risk levels and generate corresponding control instructions: when When the risk level is determined to be low, an "automatic release" control instruction is generated; when When the risk level is determined to be medium, an automatic release and additional secondary verification of specific inspection items control instruction is generated, and corresponding secondary inspection prompt information is generated and pushed to the on-site verification terminal. when When the risk level is determined to be high, a "manual key review and handling" control instruction is generated; when When the risk level is determined to be extremely high, a control command of "prohibiting entry and triggering emergency reporting" is generated, and an early warning message is automatically sent to the emergency response system.

[0041] S6: Steps for linking risk label generation with closed-loop on-site operations: The final weighted comprehensive risk scoring function F calculated in the aforementioned steps, and the complete risk assessment model formed according to the weights, thresholds, and scoring rules set by the system, after recording and confirming the validity and completeness of the scoring process and results, a comprehensive risk impact factor δ=F for identifying the risk level of vehicles entering the site is calculated, generated, and bound, or further quantified into the zoning level of freight parking management regulations according to different risk levels set by the system.

[0042] This comprehensive risk impact factor or risk level information will continue to be effective after the vehicle enters the site: The risk impact factors The data is transmitted in real time to the parking dispatch system of the chemical industrial park. The dispatch system prioritizes dispatching vehicles with high δ values ​​to sparsely populated open parking areas and locations near emergency fire lanes, with the optimization goals of "safe distance between vehicles and densely populated areas" and "facilitating rapid access to fire safety passages". The risk impact factors Real-time data is transmitted to the integrated risk monitoring and scheduling module, serving as the basis for dynamic adjustment of subsequent on-site inspection frequency—for high-risk areas… For vehicles with high quality, this module automatically increases the frame rate of AI video surveillance to increase the frequency of on-site identification and monitoring, and adjusts the on-site fixed or mobile monitoring and inspection resources. When an accident occurs in a chemical industrial park or site, the emergency command system identifies vehicles with special hazard source characteristics based on the risk impact factor δ, allocates the highest response level to emergency response resources, and provides information on hazard level and hazard mitigation strategies for response decisions.

[0043] S7: System Deployment and Operation Steps The system operates using a hybrid deployment architecture of "cloud-side business middleware + edge nodes".

[0044] One or more edge computing nodes are deployed at the entrance and exit gates and parking lot monitoring equipment of the site to complete real-time data collection, edge computing, and rapid initial risk screening. When a vehicle arrives at the site entrance, the edge nodes complete real-time data collection from S1 to S3 and cloud-based initial screening with millisecond-level response, ensuring the timeliness of temporary entry.

[0045] The cloud-based platform connects to a massive number of data sensor access points, enabling deep learning models to train on historical datasets and accident databases, and to combine weight parameters. It provides intelligent optimization and distribution updates, as well as long-term storage and business decision support for full business data and historical incident data. Simultaneously, the cloud platform receives data pushes from various edge nodes and periodically adjusts pre-device-level thresholds. Fine-tuning and calibration are performed to further improve decision-making accuracy.

[0046] In addition, edge nodes and the cloud platform are connected via an encrypted network to achieve consistent synchronization of basic configuration data and rapid issuance of commands.

[0047] Compared with the prior art, the beneficial effects of the present invention are reflected in: 1. A five-dimensional integrated quantitative evaluation framework: Existing technologies such as CN117953716A (A method for safety management and control of vehicles transporting hazardous chemicals in chemical industrial parks) focus on route planning and GPS positioning management, while CN119516752A (A method for safety supervision and alarm of vehicles transporting hazardous materials on highways) focuses on matching hazardous materials sensor data with preset standards. Neither of these technologies has comprehensively integrated entry risk factors from five dimensions: enterprise qualifications, vehicle equipment, cargo, personnel, and environment. In contrast, this invention constructs a five-in-one quantitative assessment system that integrates the qualifications of the transporting enterprise, the status of vehicle equipment, the risk of the cargo carried, the driver's status, and real-time environmental factors. It unifies and quantifies various heterogeneous parameters into comparable risk indicators, and the comprehensiveness and accuracy of risk assessment far exceed those of existing technologies.

[0048] 2. Dynamic weight adaptive adjustment mechanism: Existing risk assessment methods mostly employ static rules. The risk assessment method for hazardous chemical vehicles based on fuzzy set theory uses expert scoring to establish a membership risk assessment matrix. However, this method relies on expert experience for weight determination and lacks the ability to dynamically adjust weights according to environmental changes. The adaptive weight adjustment mechanism of this invention, through dynamic construction and real-time feedback loops, significantly outperforms existing fixed-weight risk assessment methods.

[0049] The effectiveness of data-driven technology is fully demonstrated in the access control gate testing environment, showcasing quantifiable performance advantages: Scenario 1: Effectiveness Test of Weight Adjustment under Severe Weather Conditions. In a typical severe weather scenario simulating heavy rain (visibility <100 meters), the overall vehicle risk score F value generated by the traditional fixed-weight assessment system is generally low (the average relative deviation is about 25% of the actual risk). The method of this invention dynamically increases the real-time environmental weight, which reduces the false alarm rate caused by environmental risks in the heavy rain scenario by about 48% compared with the traditional technology, and reduces the false alarm rate (high-risk vehicles are judged as low-risk and allowed to pass directly) by about 62% compared with the traditional technology.

[0050] Scenario 2: Effectiveness Test of Weight Adjustment Based on Historical Accident Time Period Analysis. In a scenario simulating late-night hours (11:00 PM to 5:00 AM the next day, historical accident statistics show that the accident rate during this period is approximately 3.5 times that of daytime), the traditional fixed-weight assessment system is too weak in considering driver state factors. The method of this invention significantly increases the driver state risk weight through adaptive adjustment. The weighting was increased by about 2.3 times compared to normal periods, which reduced the deviation between the proportion of high-risk drivers identified during this period and the actual high accident rate from about 41% in the traditional method to about 9%.

[0051] Scenario 3: Effectiveness Test of Weight Adjustment in High-Risk Goods Scenarios. In a simulated high-risk goods transportation scenario involving highly flammable liquids (closed-cup flash point, i.e., the lowest temperature at which the vapor released by the sample under specified test conditions ignites instantly upon contact with an ignition source <18℃), the traditional fixed-weight system insufficiently adjusts the risk weights for the goods, resulting in a high-risk warning omission rate of approximately 22%. The dynamic adjustment mechanism of this invention reduces this misjudgment and omission rate to approximately 4%. The above comparative test is based on six consecutive months of operational data from the chemical industrial park, with a sample size of n=12,847 vehicle trips. An A / B test was used, and the performance indicators of this method and the fixed-weight threshold method were compared concurrently. The differences were statistically significant, i.e., p<0.01.

[0052] 3. Closed-loop linkage between entry assessment and on-site operation: This invention, for the first time in a patent application, proposes a complete closed-loop mechanism linking entry assessment risk labels with parking space allocation, inspection frequency scheduling, and emergency response priority. This mechanism has been validated through long-term simulation testing based on real historical hazardous chemical parking lot operation data (sample size n=38412 vehicles, including 122 abnormal events). Using the entry risk impact factor described in this invention After implementing the parking space zoning optimization strategy, the average emergency response time for high-risk vehicles was reduced by approximately 32% to 41%. High-risk vehicles are placed in areas close to fire lanes upon entry, reducing the time for safety hazard identification and emergency response by more than 35% (due to the shortened time for parking to reach the fire lane). By actively allocating inspection resources based on the δ value, the on-site inspection coverage rate of high-risk vehicles has been increased to 99.7% (the inspection coverage rate of high-risk vehicles under the traditional solution is about 83%).

[0053] 4. Detailed and quantitative assessment of driver condition: This invention quantifies the driver's physiological state into a risk indicator that can participate in comprehensive weighted calculation through data sources such as vehicle networking and AI behavior recognition. It integrates the risk indicator with factors such as vehicle equipment and cargo physical parameters, breaking through the previous superficial control that only relied on qualification verification. This improves the control accuracy of personnel factors from the qualitative level to the quantitative assessment level. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the overall process of the intelligent entry control and risk warning method for hazardous chemical vehicles according to the present invention; Figure 2 This is a five-dimensional multi-source data fusion and risk assessment architecture diagram of a method for intelligent entry control and risk warning of hazardous chemical vehicles according to the present invention. Figure 3 This is a data flow diagram illustrating the dynamic weight adaptive adjustment mechanism of the intelligent entry control and risk warning method for hazardous chemical vehicles according to the present invention. Figure 4 This is a hierarchical control decision-making logic diagram based on multi-level early warning thresholds for a method for intelligent entry control and risk warning of hazardous chemical vehicles according to the present invention. Figure 5 This is a flowchart illustrating the entry risk label and closed-loop linkage between the on-site operation and the intelligent entry control and risk warning method for hazardous chemical vehicles according to the present invention. Detailed Implementation

[0055] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0056] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0057] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0058] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0059] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0060] It should be noted that when an element is referred to as "fixed to" or "disposed on" another element, it can be directly on the other element or there may also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "upper", "lower", "left", "right" and similar expressions used herein are for illustrative purposes only and do not represent the only implementation.

[0061] Please refer to Figures 1 to 5 For the purpose of making the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below in conjunction with specific embodiments. Embodiment 1

[0062] An actual application scenario of an intelligent entry control and risk warning method for hazardous chemical vehicles: In this embodiment, a special parking lot for hazardous chemicals in a large chemical industrial park is used as the application scenario to detail the implementation process of the method of the present invention.

[0063] In this scenario, it is identified that a hazardous chemical vehicle with the license plate number "a certain KK1** trailer" to enter the park enters the entry detection channel. The core calculation logic of the present invention is carried out by the intelligent entry control system set on the edge node server at the entrance and exit of the park.

[0064] Step 1: Multi-source data acquisition and initialization processing: When the vehicle approaches the entry detection checkpoint, the edge node control system quickly links to the following five-dimensional data acquisition tasks through the license plate recognition module and the supporting data solution, and enters the standardized processing process after recording various data in the database: (1) Carrier enterprise qualification data Acquisition: The system automatically associates the vehicle's affiliated transportation enterprise information database through the license plate recognition result, and reads in real time the licensed transportation qualification information of XX Logistics Co., Ltd. in a certain province to which the carrier enterprise "a certain KK1** trailer" belongs. The information database records that the enterprise has had 2 minor safety violations and 0 major accidents within a three-year traceability period, and the safety management system rating is B level. These original data are input into the preprocessing module.

[0065] (2) Vehicle equipment status data Data Acquisition: The system retrieves vehicle annual inspection validity information from the vehicle management database (9 months remaining, indicating basic compliance) and obtains real-time data from the onboard tire pressure sensors (average tire pressure of all four tires is 2.75 Bar, with a standard deviation of 0.12 Bar, values ​​within the normal range); braking system status parameters are read from the vehicle's central control unit, showing normal ABS status; and the onboard electrostatic grounding status is displayed as normal. Initial data is complete, meeting basic safety access requirements.

[0066] (3) Risk data of the cargo carried The system scanned the hazardous materials data in the electronic waybill. The cargo carried by the vehicle was liquid ammonia (ammonia solution), UN number 1005. The system immediately accessed the National Hazardous Chemicals Center database to retrieve the safety data sheet information: Liquid ammonia is a flammable gas at room temperature. Upon leakage, it will form an explosive mixture with air and is also toxic. It is highly susceptible to explosion and combustion upon contact with an open flame and poses corrosive and acute poisoning health hazards to humans. The hazard level is "high". The cargo volume is 24.8 tons, and the packaging method is a special vehicle-mounted liquid ammonia storage tank transport packaging that meets national standards.

[0067] (4) Driver status data Acquisition: The system acquires the driver's professional qualification information (holding a valid road dangerous goods transport qualification certificate, issued for 6 years); reads historical violation record data (accumulated points of 6 in the past 3 years, no major accidents); and performs real-time facial recognition through the vehicle-mounted AI camera. The AI ​​model judges the driver's fatigue level as "mild fatigue" (eyelid closure rate PERCLOS value is above the threshold of 0.25, which is an early fatigue warning).

[0068] (5) Real-time environmental data Data Acquisition: The system automatically acquires current environmental data via the meteorological API—the local weather is sunny, visibility is >10 km, and the road surface is dry; the urban traffic congestion index is 2.7 (mild congestion); based on the analysis of the historical accident database within the park, the peak accident period of the day is midnight.

[0069] Step 2: Multi-source data normalization processing: The edge node system calls the data processing module to sequentially map the above five dimensions of data to the [0,1] interval for quantization, classification, and normalization. Each dimension of data generates a quantized mapping relationship based on specific industry regulations and standards.

[0070] For example: Enterprise qualification data Based on the carrier's poor safety records and violation penalty records over the past three years, after normalization calculation... The initial score is 0.24; Vehicle equipment status Based on an evaluation of multiple parameters, including the remaining validity period of the vehicle's annual inspection and the real-time sensor normalization rate, the vehicle is equipped with initial... The score is 0.17; Risks of the goods carried Based on the cargo hazard classification (classification levels from lowest to highest, level 1) and the comprehensive score fitting of the transport tonnage, liquid ammonia is classified as a high-risk level (level 4-5). This cargo has a large transport tonnage and a high hazard classification. After normalized parameterization mapping... The initial score is 0.75 (this is the most prominent risk factor in this scenario); Driver status The basic risk score is 0.31, calculated by combining four indicators: driver compliance score, historical violations, and preliminary results of on-site AI fatigue detection. Real-time environmental data Real-time environmental score under conventional scoring mapping of multiple environmental factors (clear weather, good visibility) The base score is 0.12.

[0071] Step 3: Dynamic weight adaptive adjustment: Before weighting, the system reads the dynamic weight adjustment function to perform adaptive weight optimization calculation.

[0072] Considering key scene factors: It was still early, not yet nightfall, and weather conditions were favorable, but... The cargo risk is significantly high, and The fatigue detection values ​​provided an early warning of mild fatigue. Meanwhile, although the peak accident rate for the day was concentrated after midnight, the current time period is not a peak time for accidents. Therefore, the system automatically increased the cargo risk correction factor α. The weighting adjustment range, and the driver state risk weighting To improve performance under conditions such as moderate fatigue, the initial base weight values ​​for each dimension are set as follows: The basic weights are 0.10, 0.15, 0.25, 0.30, and 0.20, respectively, due to cargo risk data. An abnormally high level of risk has been identified as a core and prominent risk. Based on this, the system will... The final dynamic improvement was to 0.35 (an increase of approximately 40%). Simultaneously, based on the actual statistical distribution of risk quantification indicators across various dimensions, the weights of other dimensions were also finely adjusted. After adjustment, the total weight is normalized to 1.00, and the weight adjustments for each dimension are quickly synchronized to the algorithm calculation after meeting the operational constraints.

[0073] Step 4: Calculation of weighted risk score: The system inputs the basic risk score and the adaptively adjusted dynamic weights to calculate the comprehensive risk score:

[0074] That is, F = 0.10 × 0.24 + 0.13 × 0.17 + 0.35 × 0.75 + 0.26 × 0.31 + 0.16 × 0.12 = 0.024 + 0.0221 + 0.2625 + 0.0806 + 0.0192 = 0.4084.

[0075] The overall risk score F ultimately reached 0.4084 (range [0,1]), placing the risk level in the medium-to-high range. The system continuously monitors changes in the input data, automatically triggering a recalculation when a significant change occurs in a certain dimension of the data.

[0076] Step 5: Tiered Management and Emergency Response Plan Generation: The system stores and queries a preset risk threshold: [This is the threshold for risk thresholds]. .

[0077] Since F=0.4084, it is in to The threshold range. The system determines it to be a high-risk level, and generates a "manual key review and handling" instruction accordingly. It will push the review prompt to the on-site terminal, requesting the safety inspection personnel to conduct key verification and comparison of the high risk of leakage of flammable and toxic gases from the liquid ammonia storage tank, the record ledger and vehicle monitoring data, and to check its braking performance and the equipment of the vehicle's emergency equipment in detail.

[0078] The high-precision comprehensive risk scoring database is also linked to the comprehensive safety command center that supports the entire invention, enabling emergency reporting linkage and information sharing.

[0079] Step 6: In-depth integration and application of risk tags in on-site operations: After manual verification revealed no obvious faults, the management and security system allowed the vehicle to enter the park. Before release, the system automatically assigned a comprehensive risk impact factor label to the vehicle. This factor F is the final calculated risk score, and is further set as the risk emergency response level based on the degree of high risk.

[0080] The tag is immediately pushed to the parking space allocation and scheduling module, and the system automatically guides the vehicle to the designated parking area for high-risk vehicles near the emergency exit and away from the office area.

[0081] Meanwhile, the on-site monitoring subsystem increased the monitoring priority of "a certain KK1** trailer" based on the δ value. During its stay on-site, the frame sampling frequency of AI video monitoring was increased from 1 frame / second for normal vehicles to 2 frames / second (an increase of 100%), and the auxiliary inspection system conducted regular and close patrols of the area. The system also simultaneously pushed notices and inspection priorities for high-risk vehicles to the mobile devices of the park operations administrators.

[0082] Step 7: Emergency Response Plan and Incident Tracing When an alarm signal or other emergency occurs in the chemical industrial park, the emergency decision-making system automatically prioritizes the location of vehicles with high δ values ​​based on the risk impact factor δ, and displays the data and location information of the hazardous chemicals carried by the vehicles on the multi-mode fusion situation map of the emergency command center, so as to rationally allocate emergency response forces and improve the efficiency of emergency rescue response in accident scenarios.

[0083] Step 8: Cloud-based training and model iterative optimization The depot server pushes vehicle entry and related anomaly event tracking data to the cloud data lake daily to enrich and expand the deep learning training dataset, including the weight generator. Based on this, the security cloud side uses offline training methods to update model parameters and weight settings from time to time, enabling rapid algorithm iteration and evolution. Newly trained model versions are pushed to edge devices via OTA upgrades.

[0084] Actual running comparison test results After the park applies the method of this invention (by accessing logistics waybills for at least the first 270,000 tons of outgoing vehicles), it simultaneously tracks and statistically analyzes the overall operational safety performance. The overall time for traditional entry risk inspection of high-risk vehicles has been reduced from an average of 8.5 minutes per vehicle to 1.25 minutes, improving processing efficiency by approximately 85%. The accuracy rate of high-risk vehicle accident early warning has been improved from approximately 61% using traditional manual methods to approximately 94%. Unauthorized entry resulted in a 53.6% decrease in the accident rate inside the parking lot compared to the same period last year (based on a continuous rolling sampling of 3,000 vehicles).

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0086] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for intelligent entry control and risk warning of hazardous chemical vehicles, characterized in that, Includes the following steps: S1: Obtain multi-source data of hazardous chemical vehicles waiting to enter the site. The multi-source data includes carrier qualification data, vehicle equipment status data, cargo risk data, driver status data, and real-time environmental data. S2: Normalize the data of each dimension in S1, calculate the initial risk score of each risk feature source level, and form the initial risk score vector. S3: Based on dynamic environmental factors and real-time context information, adaptively adjust the weight allocation of each level to obtain a dynamic weight vector; S4: Calculate the overall risk value based on the weighted comprehensive risk scoring function, which is the weighted sum of the initial risk scores of each dimension and their respective dynamic weights; S5: Compare the overall risk value with the preset multi-level early warning thresholds, and generate risk control instructions of the corresponding level based on the comparison results.

2. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The carrier qualification data mentioned in step S1 includes, but is not limited to, the carrier's licensed transportation qualification information, historical safety records, number of safety violations and penalty records in the past three years, and the enterprise's safety management system rating.

3. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The vehicle equipment status data mentioned in step S1 includes basic vehicle information, vehicle annual inspection validity information, tank validity information and pressure vessel inspection certificate information, as well as real-time acquired vehicle sensor data.

4. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The cargo risk data mentioned in step S1 includes electronic waybill information, safety data sheet information for hazardous chemicals, cargo quantity and packaging method.

5. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The driver status data mentioned in step S1 includes driver qualification information, driver's license validity period, historical traffic violation records and demerit points, historical accident records, and driver physiological status data obtained in real time through the vehicle behavior detection system.

6. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The real-time environmental data mentioned in step S1 includes meteorological warning information, road condition data, real-time traffic congestion index, and statistical data on accident-prone periods based on historical data analysis.

7. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The adaptive adjustment of weight allocation at each level in step S3 includes at least one of the following rules: dynamically increasing the weight of real-time environmental data when real-time weather conditions or road conditions are worse than a preset environmental threshold; dynamically increasing the weight of driver status data when the current time period falls into a preset high-accident period; and dynamically increasing the weight of real-time environmental data and sending congestion warnings in conjunction with the vehicle navigation system when the road congestion index exceeds a preset congestion threshold. When the hazardous characteristics of the cargo are preset as high-risk substances, the weight of the risk data of the cargo is dynamically increased.

8. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The multi-level early warning thresholds mentioned in step S5 include low-risk thresholds. Medium risk threshold and high risk threshold And satisfy ; When the overall risk value When, an automatic release control command is generated; when When, an automatic release command is generated and a secondary verification control command for specific inspection items is attached; when When, generate manual key review and handling control instructions; when At that time, an entry ban will be generated and an emergency reporting and control command will be triggered.

9. The intelligent entry control and risk warning method for hazardous chemical vehicles according to claim 1, characterized in that, It also includes S6: generating a comprehensive risk impact factor based on the overall risk value calculated in step S4, binding the comprehensive risk impact factor with vehicle information and entering the site with the vehicle, and transmitting the comprehensive risk impact factor to: the parking scheduling system for optimizing the allocation of parking locations for high-risk vehicles; the comprehensive risk monitoring and scheduling module for adjusting the frequency of on-site inspections and the frequency of AI monitoring frame extraction; and the emergency command system for determining the priority of emergency response when an accident occurs.

10. The method for intelligent entry control and risk warning of hazardous chemical vehicles according to claim 1, characterized in that, The method adopts a hybrid deployment architecture of cloud + edge node. The edge node is deployed on the gate side of the site entrance and exit to realize real-time data collection and rapid risk screening, while the cloud is used to realize weight parameter optimization training and full data storage and analysis.

Images