A vehicle safety risk assessment method for SkyEye transportation management system

By using AI and natural language processing technology to build a multi-dimensional risk assessment model in road transportation, combined with blockchain technology, real-time data monitoring and dynamic adjustment are achieved, solving the problem of lack of real-time and dynamic in the existing technology, and improving the The intelligence and automation level of vehicle transportation safety management has been enhanced, and the effectiveness of risk identification and control has been enhanced.

CN118710068BActive Publication Date: 2024-11-12JIANGSU LINGHAO NETWORK TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411208865.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-12
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The existing technology lacks real-time data monitoring and dynamic adjustment mechanisms in road transportation, the system integration and feedback mechanisms are imperfect, and the risk control strategy formulation and effect prediction description are relatively general, resulting in the inadequate timely and effective assessment of vehicle safety risk.

Method used

By collecting and preprocessing multi-dimensional fusion data of vehicles, combining AI and natural language processing technology to build a risk assessment model, adopt quantitative and qualitative assessment methods, monitor and dynamically adjust risk priorities in real time, formulate and optimize risk control strategies, and use machine learning prediction measures to predict The effect is and the transparent traceability of the feedback mechanism through blockchain technology to form a closed-loop management.

Benefits of technology

It improves the intelligence and automation level of vehicle transportation safety management, enhances the accuracy and efficiency of risk identification, ensures the effectiveness and transparency of risk control measures, and achieves the flexibility and timeliness of risk management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118710068B_ABST
    Figure CN118710068B_ABST
Patent Text Reader

Abstract

The present invention discloses a vehicle safety risk assessment method for a SkyEye transportation management system, which belongs to the field of road transportation technology, and specifically includes: constructing a vehicle data set by collecting and preprocessing multi-dimensional fusion data of vehicles, combining AI and natural language processing technology; adopting a multi-dimensional risk assessment model, integrating quantitative and qualitative assessment methods, quantifying and ranking vehicle safety risks; establishing real-time data monitoring, dynamically adjusting risk priorities, formulating and optimizing risk control strategies based on risk assessment results, using machine learning to predict the effects of measures, and generating efficient execution plans; seamlessly embedding control measures into the SkyEye system, using blockchain technology to ensure that the feedback mechanism is transparent and traceable, regularly evaluating the effectiveness of strategies and intelligently optimizing them, forming a closed-loop management, and improving the intelligence and automation level of vehicle transportation safety management.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of road transportation, and specifically is a vehicle safety risk assessment method for a SkyEye transportation management system. Background Art

[0002] As the core link of the modern logistics system, vehicle road transportation is based on the use of a variety of trucks as transportation carriers to achieve efficient transfer of goods in space. In this complex production activity, the diversity of vehicle types and the variety of goods are intertwined, giving rise to a diversified logistics and transportation organization model. Each type of goods needs to be matched with the most suitable transportation vehicle and route planning according to its characteristics and destination to cope with complex and changing transportation needs. However, due to the wide range of optional routes between the starting and ending points, problems such as high vehicle empty rate and redundant mileage have become bottlenecks restricting transportation efficiency and cost control. In the face of this challenge, a series of innovative products and services aimed at optimizing vehicle-cargo matching and route planning have emerged on the market, aiming to reduce costs and increase efficiency for the freight industry through intelligent means. Although these measures have alleviated economic pressure to a certain extent, the safety risks of vehicles in road transportation cannot be ignored.

[0003] Among them, overloaded transportation is a major safety hazard in road transportation. Overloaded transportation not only accelerates the wear and tear of highway facilities, but also further reduces driving safety due to the deterioration of road conditions. On the other hand, in the whole process management of cargo transportation, there are also deficiencies in the monitoring and response measures for cargo risks. The penetration rate of modern technical means such as real-time vehicle monitoring, intelligent route optimization, and accident warning systems is not high, resulting in a lack of timely and effective response mechanisms when dealing with emergencies, which increases the uncertainty and risk of road transportation.

[0004] For example, a Chinese patent with authorization announcement number CN113065804B discloses a method and system for assessing the risk of road transportation of hazardous chemicals, including: based on the analysis of big data on the risk of hazardous chemical vehicle transportation, comprehensively identifying the risk factors that affect the safety of hazardous chemical vehicle transportation, taking the probability of possible accidents and the potential consequences of accidents in hazardous chemical vehicle transportation as the main line, using a combination of qualitative and quantitative methods to analyze the coupling correlation between multiple risk factors, and constructing estimation models for the probability of possible accidents and the potential consequences of accidents respectively. Through the analysis of the superposition effect between the two, an accurate and reliable risk assessment level for hazardous chemical vehicle transportation is obtained, thereby actively reducing the probability of possible accidents and the possible consequences of hazardous chemical vehicle transportation accidents, and improving the safety and stability of hazardous chemical vehicle transportation.

[0005] The above existing technologies all have the following problems: 1) There is no clear mention of real-time data monitoring and dynamic adjustment mechanisms, lacking real-time and dynamic characteristics; 2) The lack of system integration and feedback mechanisms; 3) They only focus on obtaining risk assessment levels, but are relatively general in terms of the formulation, implementation and effect prediction of risk control strategies. Summary of the invention

[0006] In view of the shortcomings of the existing technology, the present invention proposes a vehicle safety risk assessment method for the Sky Eye transportation management system. By collecting and preprocessing multi-dimensional fusion data of vehicles, combining AI and natural language processing technology, a vehicle data set is constructed; a multi-dimensional risk assessment model is adopted, and quantitative and qualitative assessment methods are integrated to quantify and sort vehicle safety risks; real-time data monitoring is established, and risk priorities are dynamically adjusted. Based on the risk assessment results, risk control strategies are formulated and optimized, and machine learning is used to predict the effects of measures and generate efficient execution plans; control measures are seamlessly embedded in the Sky Eye system, and blockchain technology is used to ensure that the feedback mechanism is transparent and traceable, and the effectiveness of strategies is regularly evaluated and intelligently optimized to form a closed-loop management, thereby improving the intelligence and automation level of vehicle transportation safety management.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A vehicle safety risk assessment method for a SkyEye transportation management system, comprising:

[0009] Step S1: Collect vehicle multi-dimensional fusion data, pre-process the collected vehicle multi-dimensional fusion data, use AI algorithm to identify the pre-processed vehicle multi-dimensional fusion data, and use natural language processing to analyze text data to obtain a vehicle data set;

[0010] Step S2: Construct a multi-dimensional risk assessment model, identify vehicle safety risks in combination with vehicle data sets, and use quantitative and qualitative assessment methods to quantify and score the identified risks. According to the scoring results, the risks are prioritized. At the same time, a real-time data monitoring mechanism is established to dynamically adjust the risk priority according to data changes.

[0011] Step S3: Based on the results of risk assessment and priority sorting, formulate risk control strategies, use machine learning algorithms to predict the execution effect of risk control strategies, optimize strategy combinations, and generate control measure execution plans based on the optimized strategy combinations;

[0012] Step S4: Embed control measures into the SkyEye transportation management system, establish a blockchain-based feedback mechanism, monitor and record the implementation and effects of control measures, regularly evaluate risk control strategies based on monitoring data and feedback results, and use AI algorithms to automatically optimize risk control strategies based on the evaluation results to form a closed-loop management.

[0013] Specifically, the specific steps of step S2 include:

[0014] S2.1: Determine risk assessment factors, obtain vehicle data sets, extract feature variables from vehicle data sets, and use parcel strategies to perform feature screening to obtain feature data sets , where n represents the number of feature variables in the feature data set;

[0015] S2.2: The pre-trained risk assessment model is input. The risk assessment model automatically identifies the existing security risks and their types based on the input data. For each type of risk identified, the risk assessment model performs quantitative scoring based on the risk impact and probability of occurrence, and introduces uncertainty quantification methods to evaluate the credibility interval of the scoring results.

[0016] Specifically, the specific steps of step S2 also include:

[0017] S2.3: Set the priority threshold of risk scoring, use the improved heap sorting algorithm to prioritize risks based on the scoring results, and formulate risk control and mitigation strategies based on the risk priorities;

[0018] S2.4: Collect data from vehicle operation in real time, input the real-time data into the risk assessment model, re-evaluate the risk status of the current data, introduce a dynamic adjustment mechanism, and immediately trigger the priority update process when the newly emerging risk is higher than the current highest priority risk;

[0019] S2.5: Introduce an adaptive learning mechanism and use real-time feedback data to optimize the risk assessment model.

[0020] Specifically, the specific steps of the packaging strategy in S2.1 include:

[0021] S2.11: Select the SVM base model and set the starting condition, stop condition and exhaustive search parameters of the feature search;

[0022] S2.12: Use an exhaustive search algorithm to traverse the vehicle data set to generate multiple feature subsets, use each generated feature subset to train the base model respectively, and obtain the base model evaluation result by calculating the accuracy;

[0023] S2.13: According to the evaluation results of the base model, select the feature subset with the maximum accuracy as the current optimal feature combination;

[0024] S2.14: Repeat S2.12 to S2.13 until the stopping condition is reached, and output the final selected feature set, that is, the feature combination that performs best in the entire search process.

[0025] Specifically, the specific steps of S2.2 include:

[0026] S2.21: Receive feature dataset , load the pre-trained risk assessment model and Input the pre-trained risk assessment model;

[0027] S2.22: Select a decision tree as the risk assessment model, and use historical data or simulated data to train the decision tree risk assessment model so that the decision tree risk assessment model learns the mapping relationship between risk features and risk types, and deploys the trained decision tree risk assessment model to the actual environment. When a new feature data set is input, the decision tree risk assessment model automatically performs risk assessment and outputs security risks and risk types.

[0028] Specifically, the specific steps of S2.2 also include:

[0029] S2.23: For each type of risk identified, the risk assessment model performs a quantitative score based on the risk assessment variables. The quantitative score formula is:

[0030]

[0031] in, represents the risk score, represents the weight coefficient of each type of risk, Indicates the impact of the risk. It represents the probability of risk occurrence. Indicates the risk exposure. Indicates the controllability of risk. Indicates the urgency of the risk. represents the time sensitivity factor;

[0032] S2.24: Introduce uncertainty quantification methods, build a quantitative model, and calculate the credibility interval of the scoring results based on the quantitative model. At the same time, analyze the credibility interval and evaluate the uncertainty level of the scoring results;

[0033] S2.25: Output the quantitative score and credibility interval of each risk type, and dynamically adjust the risk priority based on the score results and credibility interval.

[0034] Specifically, the specific formula of the uncertainty quantification method in S2.24 includes:

[0035]

[0036] in, represents an estimate of the uncertainty quantification result, represents the weight coefficient of the mth random sample, represents the value of the nth random variable in the mth sample, and , represents the quantized model parameters, represents the control variable parameter, represents a nonlinear function, which is and parameters , The output or result obtained after substituting into the quantitative model, M represents the number of simulations, which is the number of random samples generated, and N represents the number of random variables.

[0037] Specifically, the specific steps of the improved heap sorting algorithm in S2.3 include:

[0038] S2.31: Initialize a dynamic array for storing risk items, and set a max-heap data structure according to the risk scoring result, where each node of the max-heap is a risk item, and the value of the node is the score of the risk item;

[0039] S2.32: Add the risk items to the dynamic array in descending order according to their priorities, and build different parts of the heap in parallel, starting from the last non-leaf node of the dynamic array and building the max-heap upwards so that the value of each parent node is greater than or equal to the value of its child node;

[0040] S2.33: gradually remove the top elements of the heap and regard them as the sorted part. During the sorting process, if the score of the risk item changes, update the position of the risk item in the heap. After removing the top element of the heap, move the last element of the dynamic array to the top of the heap, and adjust the heap from top to bottom and from left to right.

[0041] S2.34: Repeat S2.33 until all elements are removed and sorted;

[0042] S2.35: During the sorting process, the interface or log is updated in real time, and after the sorting is completed, an interactive interface or API is provided to output the sorted risk item list to the user in the form of a list.

[0043] Specifically, the risk control strategy in step S3 includes: strengthening vehicle maintenance, improving driver training, optimizing driving routes, and installing safety equipment.

[0044] Specifically, an electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a vehicle safety risk assessment method for a SkyEye transportation management system when executing the computer program.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The present invention proposes a vehicle safety risk assessment method for the SkyEye transportation management system. By collecting multi-dimensional fusion data of vehicles and performing preprocessing, combined with AI algorithms and natural language processing technology, the potential safety risks of vehicles can be more accurately identified, thereby improving the accuracy and efficiency of risk identification. A multi-dimensional risk assessment model is constructed to not only identify risks, but also quantitatively score risks through quantitative and qualitative assessment methods, and sort them according to the scoring results, which helps decision makers to quickly focus on high-risk areas. At the same time, the real-time data monitoring mechanism ensures the dynamic and timeliness of risk assessment, making risk management more flexible and effective.

[0047] 2. The present invention proposes a vehicle safety risk assessment method for the SkyEye transportation management system. Based on the risk assessment results, a targeted risk control strategy is formulated, and the effects of control measures are predicted through a machine learning algorithm to further optimize the strategy combination. This data-driven method improves the effectiveness and targeting of risk control measures and reduces unnecessary waste of resources; it realizes the automation and intelligence of risk management, and the introduction of blockchain technology ensures the security and transparency of the feedback mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of a vehicle safety risk assessment method of a SkyEye transportation management system of the present invention;

[0049] Figure 2 This is a principle flow chart of a vehicle safety risk assessment method of a SkyEye transportation management system of the present invention;

[0050] Figure 3 A risk assessment flow chart of a vehicle safety risk assessment method of a SkyEye transportation management system of the present invention;

[0051] Figure 4 This is a flow chart of a heap sorting algorithm implementation of a vehicle safety risk assessment method for a SkyEye transportation management system of the present invention. DETAILED DESCRIPTION

[0052] Example 1

[0053] See also Figure 1-Figure 2 , an embodiment of the present invention provides: a vehicle safety risk assessment method for a SkyEye transportation management system, comprising the following steps:

[0054] Step S1: Collect vehicle multi-dimensional fusion data, pre-process the collected vehicle multi-dimensional fusion data, use AI algorithm to identify the pre-processed vehicle multi-dimensional fusion data, and apply natural language processing to analyze text data to obtain a vehicle data set;

[0055] Among them, vehicle multi-dimensional fusion data includes: basic vehicle information, transportation environment data, driver violation records, historical accident data, and transportation process data.

[0056] Step S2: Construct a multi-dimensional risk assessment model, identify vehicle safety risks in combination with vehicle data sets, and use quantitative and qualitative assessment methods to quantify and score the identified risks. According to the scoring results, the risks are prioritized. At the same time, a real-time data monitoring mechanism is established to dynamically adjust the risk priority according to data changes.

[0057] Among them, vehicle safety risk factors include: vehicle failure, driver fatigue, illegal driving, road conditions, and environmental factors.

[0058] Step S3: Based on the results of risk assessment and priority sorting, formulate risk control strategies, use machine learning algorithms to predict the execution effect of risk control strategies, optimize strategy combinations, and generate control measure execution plans based on the optimized strategy combinations;

[0059] Among them, risk control strategies include: strengthening vehicle maintenance, improving driver training, optimizing driving routes, and installing safety equipment.

[0060] Step S4: Embed control measures into the SkyEye transportation management system, establish a blockchain-based feedback mechanism, monitor and record the implementation and effects of control measures, regularly evaluate risk control strategies based on monitoring data and feedback results, and use AI algorithms to automatically optimize risk control strategies based on the evaluation results to form a closed-loop management.

[0061] Example 2

[0062] See also Figure 3-Figure 4 In this embodiment, the specific steps of step S2 include:

[0063] S2.1: Determine risk assessment factors, obtain vehicle data sets, extract feature variables from vehicle data sets, and use parcel strategies to perform feature screening to obtain feature data sets , where n represents the number of feature variables in the feature data set;

[0064] Furthermore, the specific steps of extracting feature variables from the vehicle data set include:

[0065] (1) Identify risk factors, such as vehicle failure, driver fatigue, illegal driving, road conditions, and environmental factors;

[0066] (2) Dataset exploration:

[0067] 1) Check the overall structure of the vehicle dataset, including data fields, data types, and data ranges;

[0068] 2) Analyze the distribution of each data field to understand the statistical characteristics of the data's central tendency and degree of dispersion;

[0069] 3) Check missing values ​​in the data and fill or delete them;

[0070] (3) Feature selection and correlation analysis:

[0071] 1) Based on business knowledge and experience, preliminarily screen out characteristic variables related to risk factors;

[0072] 2) Using the statistical method of Pearson's correlation coefficient to quantify the correlation between the characteristic variables and the risk factors, wherein the calculation of the Pearson's correlation coefficient is a prior art in the art and is not an inventive solution of the present application, and will not be described in detail here;

[0073] 3) According to the results of correlation analysis, filter out the characteristic variables that are irrelevant or weakly correlated with risk factors, and sort the remaining characteristic variables according to the strength of correlation;

[0074] (4) Feature Engineering:

[0075] 1) Perform necessary transformations on the selected feature variables, such as standardization, normalization, and encoding, to improve the performance of the model;

[0076] 2) Combine multiple feature variables into new features to capture more complex relationships and patterns;

[0077] (5) Feature verification:

[0078] 1) Use cross-validation techniques to evaluate the impact of different feature combinations on model performance;

[0079] 2) During the model training process, the importance of each feature is evaluated to further confirm the effectiveness and necessity of the selected features.

[0080] S2.2: Input the pre-trained risk assessment model. The risk assessment model automatically identifies the existing security risks and their types based on the input data. For each type of risk identified, the risk assessment model quantifies the risk impact and probability of occurrence, and introduces uncertainty quantification methods to evaluate the credibility interval of the scoring results.

[0081] S2.3: Set the priority threshold of risk scoring, use the improved heap sorting algorithm to prioritize risks based on the scoring results, and formulate risk control and mitigation strategies based on the risk priorities;

[0082] S2.4: Collect data from vehicle operation in real time, input the real-time data into the risk assessment model, re-evaluate the risk status of the current data, introduce a dynamic adjustment mechanism, and immediately trigger the priority update process when the newly emerging risk is higher than the current highest priority risk;

[0083] S2.5: Introduce an adaptive learning mechanism and use real-time feedback data to optimize the risk assessment model.

[0084] The specific steps of the package strategy in S2.1 include:

[0085] S2.11: Select the SVM base model and set the starting condition, stop condition and exhaustive search parameters of the feature search;

[0086] S2.12: Use an exhaustive search algorithm to traverse the vehicle data set to generate multiple feature subsets, use each generated feature subset to train the base model respectively, and obtain the base model evaluation result by calculating the accuracy;

[0087] Furthermore, the exhaustive search algorithm in the present invention selects depth-first search, which has relatively low space complexity and is suitable for situations where the solution space is large. The core of depth-first search is: starting from the initial node, going deeper along a path until a leaf node is reached or it is impossible to go deeper, then backtracking to the previous node and selecting another path to continue searching.

[0088] S2.13: According to the evaluation results of the base model, select the feature subset with the maximum accuracy as the current optimal feature combination;

[0089] S2.14: Repeat S2.12 to S2.13 until the stopping condition is reached, and output the final selected feature set, that is, the feature combination that performs best in the entire search process.

[0090] The specific steps of S2.2 include:

[0091] S2.21: Receive feature dataset , load the pre-trained risk assessment model and Input the pre-trained risk assessment model;

[0092] S2.22: Select a decision tree as the risk assessment model, use historical data or simulated data to train the decision tree risk assessment model, so that the decision tree risk assessment model learns the mapping relationship between risk features and risk types, and deploys the trained decision tree risk assessment model to the actual environment. When a new feature data set is input, the decision tree risk assessment model automatically performs risk assessment and outputs security risks and risk types;

[0093] S2.23: For each type of risk identified, the risk assessment model performs a quantitative score based on the risk assessment variables. The quantitative score formula is:

[0094]

[0095] in, represents the risk score, The weight coefficient for each type of risk is used to adjust the relative importance of each parameter in the overall risk score, which is determined by expert evaluation, historical data analysis, or machine learning algorithms. It indicates the impact of risk. It is a quantitative score that reflects the possible loss or damage after the risk event occurs. It is set according to business rules or expert opinions. It indicates the probability of risk occurrence and is also a quantitative score, which is predicted by the model based on historical data and real-time data. Indicates the risk exposure, that is, the degree to which a system, asset or person is at risk. Indicates the controllability of risk, using its reciprocal, because the higher the controllability, that is, the smaller the value, the lower the risk score should be. If the risk is completely uncontrollable, the value of this item will be very large, thereby increasing the total risk score. Indicates the urgency of the risk, that is, whether the risk is about to occur or has already occurred. It is mainly quantified based on the current status or forecast of the risk event. It represents the time sensitivity factor, which is used to consider the change of risk over time. It is a function of time t, reflecting the increase or decrease of risk over time.

[0096] S2.24: Introduce uncertainty quantification methods, build quantitative models, and calculate the credibility interval of the scoring results based on the quantitative models. At the same time, analyze the credibility interval and evaluate the uncertainty level of the scoring results. The specific formulas of the uncertainty quantification method include:

[0097]

[0098] in, represents an estimate of the uncertainty quantification result, Represents the weight coefficient of the mth random sample, which is used to reflect the different impacts of the sample on the final estimated value. represents the value of the nth random variable in the mth sample, and , represents the quantized model parameters, represents the control variable parameter, represents a nonlinear function, which is and parameters , The output or result obtained after substituting into the quantitative model, M represents the number of simulations, which is the number of random samples generated. The larger M is, the closer the estimated value is to the true value. N represents the number of random variables. Used to normalize the weight coefficients to ensure that the estimated value is not affected by the sum of the weights;

[0099] Furthermore, the calculation formula of the credibility interval is:

[0100]

[0101] Where Q represents the confidence interval, represents the weighted estimate of the overall mean, k represents the number of independent sample groups, represents the weighted variance, represents the upper bound of the z distribution with degrees of freedom df Quantile, represents the sample size of the i-th group of samples, Represents the weighted adjustment of the variance estimate of each group of samples, taking into account the impact of the sample size on the variance estimate. represents the weight of the i-th group of independent samples, It represents the adjustment of the width of the confidence interval, taking into account the impact of different group sample sizes and weights on the confidence interval.

[0102] In addition, the specific steps of the uncertainty quantification method include:

[0103] (1) Determine the model and parameters: Identify the sources of uncertainty and related random variables that need to be evaluated, and determine the probability distribution type and parameters of the random variables, such as normal distribution and uniform distribution;

[0104] (2) Generate random samples: Use the random number generator method to generate a large number of random samples based on the probability distribution of random variables;

[0105] (3) Simulation experiment: For each random sample, perform calculations or simulation experiments based on the problem model to obtain the corresponding output or results;

[0106] (4) Statistical analysis: Statistical analysis is performed on the results of all simulation experiments, and the statistics of sample mean, standard deviation, and confidence interval are calculated to estimate the distribution and uncertainty range of the target variable.

[0107] S2.25: Output the quantitative score and credibility interval of each risk type, and dynamically adjust the risk priority based on the score results and credibility interval.

[0108] The specific steps of the improved heap sort algorithm in S2.3 include:

[0109] S2.31: Initialize a dynamic array for storing risk items, and set a max-heap data structure according to the risk scoring result, where each node of the max-heap is a risk item, and the value of the node is the score of the risk item;

[0110] S2.32: Add the risk items to the dynamic array in descending order according to their priorities, and build different parts of the heap in parallel, starting from the last non-leaf node of the dynamic array and building the max-heap upwards so that the value of each parent node is greater than or equal to the value of its child node;

[0111] S2.33: gradually remove the top elements of the heap and regard them as the sorted part. During the sorting process, if the score of the risk item changes, update the position of the risk item in the heap. After removing the top element of the heap, move the last element of the dynamic array to the top of the heap, and adjust the heap from top to bottom and from left to right.

[0112] S2.34: Repeat S2.33 until all elements are removed and sorted;

[0113] S2.35: During the sorting process, the interface or log is updated in real time, and after the sorting is completed, an interactive interface or API is provided to output the sorted risk item list to the user in the form of a list.

[0114] The risk control strategies in step S3 include: strengthening vehicle maintenance, improving driver training, optimizing driving routes, and installing safety equipment.

[0115] Example 3

[0116] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of a vehicle safety risk assessment method for a SkyEye transportation management system are implemented. For details, please refer to the above method embodiment, which will not be repeated here.

[0117] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention, and all of these are within the protection of the present invention.

Claims

1. A vehicle safety risk assessment method for a SkyEye transportation management system, characterized in that: include: Step S1: Collect vehicle multi-dimensional fusion data, pre-process the collected vehicle multi-dimensional fusion data, use AI algorithm to identify the pre-processed vehicle multi-dimensional fusion data, and use natural language processing to analyze text data to obtain a vehicle data set; The multi-dimensional fusion data of vehicles includes: basic vehicle information, transportation environment data, driver violation records, historical accident data, and transportation process data; Step S2: Construct a multi-dimensional risk assessment model, identify vehicle safety risks in combination with vehicle data sets, and use quantitative and qualitative assessment methods to quantify and score the identified risks. According to the scoring results, the risks are prioritized. At the same time, a real-time data monitoring mechanism is established to dynamically adjust the risk priority according to data changes. Step S3: Based on the results of risk assessment and priority sorting, formulate risk control strategies, use machine learning algorithms to predict the execution effect of risk control strategies, optimize strategy combinations, and generate control measure execution plans based on the optimized strategy combinations; Step S4: Embed the control measures into the SkyEye transportation management system, establish a feedback mechanism based on blockchain, monitor and record the implementation and effects of the control measures, regularly evaluate the risk control strategy based on the monitoring data and feedback results, and automatically optimize the risk control strategy using AI algorithms based on the evaluation results to form a closed-loop management; The specific steps of step S2 include: S2.1: Determine risk assessment factors, obtain vehicle data sets, extract feature variables from vehicle data sets, and use parcel strategies to perform feature screening to obtain feature data sets , where n represents the number of feature variables in the feature data set; S2.2: Input the pre-trained risk assessment model. The risk assessment model automatically identifies the existing security risks and their types based on the input data. For each type of risk identified, the risk assessment model quantifies the risk impact and probability of occurrence, and introduces uncertainty quantification methods to evaluate the credibility interval of the scoring results. The specific steps of step S2 also include: S2.3: Set the priority threshold of risk scoring, use the improved heap sorting algorithm to prioritize risks based on the scoring results, and formulate risk control and mitigation strategies based on the risk priorities; S2.4: Collect data from vehicle operation in real time, input the real-time data into the risk assessment model, re-evaluate the risk status of the current data, introduce a dynamic adjustment mechanism, and immediately trigger the priority update process when the newly emerging risk is higher than the current highest priority risk; S2.5: Introduce an adaptive learning mechanism and use real-time feedback data to optimize the risk assessment model.

2. The vehicle safety risk assessment method of the SkyEye transportation management system according to claim 1 is characterized in that: The specific steps of the packaging strategy in S2.1 include: S2.11: Select the SVM base model and set the starting condition, stop condition and exhaustive search parameters of the feature search; S2.12: Use an exhaustive search algorithm to traverse the vehicle data set to generate multiple feature subsets, use each generated feature subset to train the base model respectively, and obtain the base model evaluation result by calculating the accuracy; S2.13: According to the evaluation results of the base model, select the feature subset with the maximum accuracy as the current optimal feature combination; S2.14: Repeat S2.12 to S2.13 until the stopping condition is reached, and output the final selected feature set, that is, the feature combination that performs best in the entire search process.

3. The vehicle safety risk assessment method of the SkyEye transportation management system according to claim 2 is characterized in that: The specific steps of S2.2 include: S2.21: Receive feature dataset , load the pre-trained risk assessment model and Input the pre-trained risk assessment model; S2.22: Select a decision tree as the risk assessment model, and use historical data or simulated data to train the decision tree risk assessment model so that the decision tree risk assessment model learns the mapping relationship between risk features and risk types, and deploys the trained decision tree risk assessment model to the actual environment. When a new feature data set is input, the decision tree risk assessment model automatically performs risk assessment and outputs security risks and risk types.

4. The vehicle safety risk assessment method of the SkyEye transportation management system as claimed in claim 3 is characterized in that: The specific steps of S2.2 also include: S2.23: For each type of risk identified, the risk assessment model performs a quantitative score based on the risk assessment variables. The quantitative score formula is: ; in, represents the risk score, represents the weight coefficient of each type of risk, Indicates the impact of the risk. It represents the probability of risk occurrence. Indicates the risk exposure. Indicates the controllability of risk. Indicates the urgency of the risk. represents the time sensitivity factor; S2.24: Introduce uncertainty quantification methods, build a quantitative model, and calculate the credibility interval of the scoring results based on the quantitative model. At the same time, analyze the credibility interval and evaluate the uncertainty level of the scoring results; S2.25: Output the quantitative score and credibility interval of each risk type, and dynamically adjust the risk priority based on the score results and credibility interval.

5. The vehicle safety risk assessment method of the SkyEye transportation management system as claimed in claim 4 is characterized in that: The specific formula of the uncertainty quantification method in S2.24 includes: ; in, represents an estimate of the uncertainty quantification result, represents the weight coefficient of the mth random sample, represents the value of the nth random variable in the mth sample, and , represents the quantized model parameters, represents the control variable parameter, represents a nonlinear function, which is and parameters , The output or result obtained after substituting into the quantitative model, M represents the number of simulations, which is the number of random samples generated, and N represents the number of random variables.

6. The vehicle safety risk assessment method of the SkyEye transportation management system according to claim 5 is characterized in that: The specific steps of the improved heap sorting algorithm in S2.3 include: S2.31: Initialize a dynamic array for storing risk items, and set a max-heap data structure according to the risk scoring result, where each node of the max-heap is a risk item, and the value of the node is the score of the risk item; S2.32: Add the risk items to the dynamic array in descending order according to their priorities, and build different parts of the heap in parallel, starting from the last non-leaf node of the dynamic array and building the max-heap upwards so that the value of each parent node is greater than or equal to the value of its child node; S2.33: gradually remove the top elements of the heap and regard them as the sorted part. During the sorting process, if the score of the risk item changes, update the position of the risk item in the heap. After removing the top element of the heap, move the last element of the dynamic array to the top of the heap, and adjust the heap from top to bottom and from left to right. S2.34: Repeat S2.33 until all elements are removed and sorted; S2.35: During the sorting process, the interface or log is updated in real time, and after the sorting is completed, an interactive interface or API is provided to output the sorted risk item list to the user in the form of a list.

7. The vehicle safety risk assessment method of the SkyEye transportation management system according to claim 6 is characterized in that: The risk control strategies in step S3 include: strengthening vehicle maintenance, improving driver training, optimizing driving routes, and installing safety equipment.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, it implements the steps of a vehicle safety risk assessment method for a SkyEye transportation management system as described in any one of claims 1-7.