Research On Vehicle Speeed And Driving Safety Based On Data Characteristics

Safety is an eternal subject in the research of vehicle operation control and driving behavior recognition.The driving behavior of vehicle drivers is becoming more and more personalized and complicated,and the individual correlation between the safe operation state of vehicle and the driving behavior of vehicle drivers is becoming more and more strong.The road traffic accidents caused by this cause massive loss to personal property and social economy,and increase the operating cost of transportation system.In order to more scientifically and objectively reflect the real running state of vehicles,especially the safety evaluation of vehicle driving speed,it is necessary to study new evaluation methods to achieve the description,discrimination and diagnosis of safety characteristics of driving speed.In this paper,typical traffic accidents in a city and the driving data of common M1 passenger vehicles are selected as samples.Based on the theoretical relationship between risk and hazard,the relevant factors of driving speed control are discussed and the main causes of traffic accidents are analyzed.Vehicle diagnosis system and specific data processing tools were used to collect natural driving state data,and data mining methods were used to select and construct vehicle speed state safety assessment indicators.The correlation analysis Apriori algorithm is applied to analyze the data of a single driving segment,and the correlation characteristics of vehicle speed control behavior are obtained.In view of the limitations of traditional safety driving evaluation model,support vector machine(SVM,RF,XGBoost,etc.)analysis and decision model is proposed to judge the safety degree of driver’s control of vehicle speed state in fixed time period and road section.The main influencing factors of vehicle speed control are obtained by means of K-means clustering analysis and decision tree classification method.Based on the above research,the main results are as follows:(1)In order to clarify the influence of vehicle speed on vehicle safety,based on risk and hazard theory,the related factors of vehicle speed control were discussed,which laid a foundation for vehicle speed control,analysis and evaluation.Analysis of the causes of traffic accidents in a city,in-depth analysis of human factors,among which the accidents caused by"speed control problems" accounted for 41.7%,the accidents caused by "improper driving behavior" accounted for 37.5%,the accidents caused by "do not obey the traffic rules"accounted for 20.8%.Aiming at the braking performance,one of the key indexes affecting the safety of vehicle operation,the braking performance test of snow and ice pavement was carried out in combination with the operating conditions of snow and ice pavement in winter of a city.Braking tests were carried out on test vehicles at speeds of 10km/h,20km/h,30km/h,40km/h,50km/h and 60km/h,respectively,on compacted snow and ice covered surface and non-snow and ice covered surface with 48mm thickness.The relationship between vehicle driving speed and braking time,and between vehicle driving speed and braking distance was analyzed.The test results show that the minimum braking distance is 18.36m on the road without snow and ice cover,and the maximum braking distance is 94.59m on the road with snow and ice cover under the driving speed level of 60km/h.The randomness of the relationship between braking distance and vehicle speed on snow and ice surface increases the risk of vehicle running on snow and ice surface.By analyzing the mean difference,standard deviation and coefficient of variation of the experimental data of all groups,the degree of dispersion of the experimental data of each group was further determined.Each data of discrete degree increased with the increasing of speed,braking distance and braking time of standard deviation of the snow and ice road(mean difference)were higher than the ice and snow road,braking distance is significantly influenced by the initial braking speed,and presents the positive correlation,this shows that when the car speed increase,the braking time and braking distance become unstable.State control of vehicle driving speed is very important to driving safety.(2)In order to obtain the driving state data,OBD is used to collect driving state data.The data collection environment takes into account vehicle type,displacement,driver gender and age,as well as seasonal factors.1.16×106 valid data most closely related to driver’s control behavior were collected,and the data were screened and cleaned.Safety threshold and dangerous critical value were set,and then the driver’s driving behavior was analyzed.Z-score standardization method was used to standardize the coded data,and KMO(Kaiser-MeyerOlkin)correlation test was carried out on the standardized data.Preliminary conception of data pretreatment was put forward to prepare for the establishment of evaluation indicators.The paper selects and processes 7 parameters,such as average speed,over speed,sharp acceleration and sharp deceleration,as the indicators of vehicle speed state safety evaluation,and analyzes and evaluates the safety of representative drivers’ speed state.(3)In order to obtain the internal logical relationship between the vehicle operation data,the correlation characteristics of driving speed control are analyzed.Through the analysis,the association rules are obtained,that is,the frequency of the occurrence of key parameter indexes is obtained based on the analysis results of test data.Based on association rule Apriori method,four strong correlation rules(80%-100%),two middle correlation rules(50%-80%)and two weak correlation rules(0-50%)were obtained.(4)In order to deeply analyze the safe driving behavior of different drivers,the acceleration ratio,deceleration ratio,average speed and other parameters are taken as data mining parameter variables.By means of K-means clustering analysis and decision tree classification method,the clustering analysis of acceleration ratio,deceleration ratio,average speed and other indicators is carried out,and it is concluded that the ratio of acceleration and deceleration has a significant impact on driving safety.When the ratio of acceleration and deceleration is large,there are more variable speed operations in driving process,which increases the cost of vehicle safety maintenance and energy consumption.Based on the data of vehicle speed and acceleration,the driving speed control behavior of drivers is defined and evaluated.Considering vehicle operation data,road conditions,climate conditions and other factors,the visualized image directly reflects each driver’s speed control behavior and the probability of unsafe speed,acceleration and other indicators,which provides an important basis for the safety assessment of different drivers’ individual driving speed state.(5)Using machine learning decision model,the vehicle driving safety was evaluated by integrating the vehicle speed,acceleration and other parameters in urban roads.Based on the pattern recognition method of SVM,the classification effect of safe driving,risky driving and dangerous driving is finally obtained.The same evaluation method was adopted as SVM method,and RF and XGBoost test were analyzed and evaluated by training model.It can be seen from the results of SVM,RF and XGBoost prediction models for driving behavior that the prediction accuracy of the three prediction models is relatively excellent.In order to improve the traffic safety of urban roads,the identification algorithms of various dangerous driving behavior indicators are established for the possible problems of dangerous driving behavior.Based on the test data,the driver’s style was classified by K-means clustering algorithm,and the trained model was verified and tested.Finally,the ideal accuracy rate and the ideal recognition effect were obtained.Based on vehicle driving state data,the paper focus on safe driving behavior,quantitative assessment of drivers the security of the state of the individual to the car speed control,as a risk evaluation index,a method for building the micro driving safety system foundation,provide a reference for driving training and safety awareness.It can also provide decisionmaking support for management departments to formulate relevant provisions and regulations.