Research On Driving Risk Recognition And Safety Evaluation Of Commercial Vehicle Based On Data Mining Technology

In recent years,with the vigorous development of China’s logistics industry,highway transportation has become more frequent,resulting in a high freight accident rate.The safety problem in the transportation process has become a hot issue to be solved.According to statistics,75%of traffic accidents are caused by the improper operation of the driver.Therefore,the study of driver behavior is of great significance to ensure vehicle driving safety and transportation safety.Based on big data of commercial vehicle networking,this paper uses data mining technology to mine the driving laws behind the big data,analyze the driving behaviors of drivers,study the driving risk identification of commercial vehicles,and establish a quantitative safety evaluation system.The main contents are as follows:(1)Data acquisition,preprocessing and feature extraction for commercial vehicle networking.CAN bus technology,high-precision sensors and GPS global positioning system are used to collect natural driving data from commercial vehicles.The travel data of 50 drivers within one month is extracted from the vehicle operation monitoring platform and preprocessed,including trips division,short trips removal,missing data removal and interpolation,outlier detection,data smoothing filtering,etc.Overall consideration the influence of driver factors(bad driving behavior operation,etc.),vehicle driving state(driving speed,driving acceleration,etc.)and road environmental factors(daytime,nighttime,etc.)on the driving safety state.Eighteen characteristic parameters that can characterize the driving risk of commercial vehicles are extracted and constructed,and the data dimension reduction and optimization are carried out,which are used as technical reserves for subsequent research.(2)Commercial vehicle driving risk level identification.In view of the current problems of unreasonable classification of driving risks and low identification accuracy,a commercial vehicle driving risk identification model based on the Extreme Gradient Boosting(XGBoost)algorithm is established.The data mining technology is used to classify and cluster the driving risks of commercial vehicles,and the drivers with high driving risk tendencies are accurately captured.A more powerful and faster-solving XGBoost algorithm is used to establish a driving risk identification model,which is compared with traditional integrated learning algorithms(decision tree,random forest,K-nearest neighbor,etc.).The experimental results show that the model has advantages in recognition accuracy and can effectively suppress the imbalance of samples.(3)Safety quantitative evaluation of commercial vehicle driving behavior.The purpose of this study is to evaluate the driving safety of commercial vehicles quantitatively.The driving safety of different drivers is evaluated based on data-driven technology.Combining the entropy weight method(EWM),the analytic hierarchy process(AHP),and the fuzzy comprehensive evaluation algorithm,a comprehensive evaluation model of commercial vehicle driving behavior safety based on multi-membership function is proposed.Firstly,the safety evaluation system of commercial vehicle driving behavior is established,and the traditional entropy weight-AHP(EW-AHP)is improved to comprehensively empower each evaluation index and determine the key factors affecting driving safety.Then the fuzzy comprehensive evaluation matrix is calculated by combining multi-membership function and fuzzy mathematics theory to realize the quantitative evaluation of driving behavior safety.Finally,the rationality and reliability of the model are verified by real vehicle data.(4)The design and implementation of a commercial vehicle driving risk identification and safety evaluation system.The algorithm model is applied to practice,a software system is developed which can monitor the running state of commercial vehicles in real time,identified driving risks and evaluated driving safety quantitatively.In this study,a driver’s "driving-evaluation-improvement-driving" closed-loop feedback driving behavior optimization model is realized,which helps the driver to improve driving habits and improve driving safety.