Research On Real-time Safety Evaluation Method Of Driving Behavior Based On Improved TOPSIS

With the rapid growth of the number of motor vehicles in my country and the rapid development of the transportation industry,the property losses and casualties caused by road traffic accidents have a huge impact on people’s normal life.How to effectively reduce traffic accidents and improve road traffic safety has become a major problem.In the road traffic system,the driver is the main body,and the behavior of the driver is closely related to road traffic safety;in the same road traffic environment,the driving habits of different drivers are quite different,and the risks of accidents faced by different drivers are also different.the same.Therefore,it is necessary to analyze the behavior characteristics of different drivers and construct an evaluation index system to evaluate the safety of drivers’ behavior.Based on the current research on driving behavior in various countries,it can be found that the existing research is not complete enough for the experimental design of relevant data collection in the driving process of drivers,and the research on the analysis and modeling of driving behavior safety evaluation indicators is also lacking.The safety evaluation index system of driving behavior is not complete enough,and the safety evaluation is not accurate enough.Taking into account the shortcomings of current research,this paper proposes a method for real-time safety assessment of driving behavior based on improved TOPSIS.The main contents are as follows:(1)Firstly,analyze from three aspects,and initially establish a driving behavior safety evaluation index system.The safety evaluation index system includes the following ten elements: the safety of visual perception behavior when changing lanes,the safety of visual perception of obstacles ahead,and the control Dynamic amplitude change rate,steering wheel angle entropy value,steering wheel angle standard deviation,lateral acceleration standard deviation,lateral lane keeping safety,longitudinal speed standard deviation,longitudinal acceleration standard deviation,and longitudinal acceleration change rate.(2)In order to more accurately collect the relevant data generated by the driver during driving and calculate the relevant indicators,a real-vehicle test method was designed,and 12 drivers were recruited to participate in the data collection.The K-means method is used to perform initial clustering on the driver’s gaze point data collected by the eye tracker,and use the principle to screen out some abnormal points and eliminate them,so as to obtain a driver’s gaze point distribution map that is more conducive to analysis,and further analyze it.The safety index of the visual perception behavior of the driver when changing lanes.(3)In order to reduce the computational complexity and avoid index redundancy,the Pearson correlation analysis is used to analyze the 10 safety evaluation index data of12 drivers initially determined,and remove the more relevant similar indexes(longitudinal acceleration standard).Poor)to obtain 9 key indicators for safety evaluation including steering wheel angle entropy.Then the AHP-Entropy method combined subjective and objective is used to determine the weight of key indicators to make the obtained indicator weights more reasonable.(4)Improve the traditional TOPSIS multi-attribute decision-making method,combine the unique advantages of Mahalanobis distance and contact vector distance,replace the Euclidean distance in the original method,and recalculate the new relative closeness of each driver to the positive and negative ideal solution.According to the calculation results,the safety levels of the 12 drivers’ behaviors were sorted.After a comparative analysis with the traditional method,it was found that the driving behavior safety level ranking results of the improved method were more accurate,which verified the feasibility and accuracy of the proposed improved method.The improved TOPSIS driving behavior real-time safety assessment method proposed in this paper can help improve the driver’s behavioral safety awareness,promptly detect and correct bad driving habits,and reduce safety hazards;it provides theoretical support for the development of vehicle advanced driving assistance systems.