Research On Driving Safety Of Heavy Trucks On Expressway Based On Driving Style

In recent years,with the rapid development of the logistics industry and increasing freight demand,heavy trucks have become the primary mode of goods transportation.However,due to the continuous rise in expressway traffic volume,the accident rate remains high.Traffic accidents caused by heavy trucks are often serious,causing widespread concern in society.Statistics show that more than 90% of traffic accidents are caused by improper driver operation.Therefore,studying the driving safety issues of heavy truck drivers is of great significance.Based on the dataset of heavy trucks,this paper focuses on analyzing drivers’ driving patterns on expressways,identifying drivers’ driving styles,predicting their abnormal driving behaviors,and proposing safe driving recommendations.The main objectives of this paper can be summarized as follows:(1)Acquiring and pre-processing heavy truck driving data to construct an expressway driving dataset.Firstly,actual heavy truck driving data from Vehicular Ad Hoc Networks(VANETs)was accessed,and data from 12 drivers’ driving over a onemonth period were selected.Then,criteria for road segment construction were defined,and the AMAP API was used to filter out the expressway data for each truck.The data was subsequently pre-processed using common methods such as data cleaning,and 14 feature parameters were selected to construct the expressway driving dataset.The use of road segments improved the granularity of data cleaning,resulting in a significant improvement in the quality of the pre-processed heavy truck data.(2)Using the k-means++ algorithm in conjunction with ant colony optimization to classify the driving styles of heavy trucks and build an SVM model for identifying drivers’ driving styles.Firstly,21 feature parameters related to speed,acceleration,and speeding were selected to build the driving style dataset.Secondly,factor analysis was used to reduce the dimensionality of the data.The ant colony algorithm,known for finding global optimal characteristics,was used to find the optimal initial cluster center for the k-means++ algorithm.Subsequently,k-means++ was used to cluster the data,resulting in heavy truck drivers being divided into two driving styles: adventurous and conservative.Finally,an SVM model was built to identify the online driving style of heavy truck drivers,with experimental results showing a model recognition ratio accuracy rate of 0.97.(3)The third objective is to predict drivers’ abnormal driving behavior in real time,considering different driving styles.In this paper,12 feature parameters that can express driving behavior were selected based on factors related to the driver’s driving and the truck’s power,and a driving behavior prediction dataset was constructed.The abnormal discrimination criteria were designed using the interquartile distance of the error between the predicted value and the true value.A prediction model based on recurrent neural network was used to forecast abnormal driving behavior in advance.Experimental results for adventurous and conservative driving styles showed that the proportion of abnormal driving behavior among adventurous drivers was approximately2% higher than that among conservative drivers.