Research On Short-term Driving Style Evaluation Method Based On Driving Proneness

There are varying degrees of differences in the driving styles of different drivers,which can generally be divided into aggressive,neutral,and conservative types.However,driving style is affected by factors such as driving tasks,driving environment and changes between different styles often occur,especially changes in specific scenarios are highly random.When traffic conflicts occur between multiple parties,the quantitative evaluation results of short-term driving styles are an important basis for formulating early warning information and decision-making suggestions.If the driver in traffic conflicts can timely understand the other’s driving behavior characteristics and take targeted actions,then the conflicts can be resolved,which can greatly improve the road safety level and traffic capacity.The characteristics of the short-term driving style are mainly reflected in the differences between styles in different scenarios,and the evaluation results of styles need to be updated in real time based on the short-term data.Based on the above characteristics,the concept of driving proneness was proposed as a bridge connecting specific operations and short-term driving styles in this paper.The driving proneness was used as an evaluation tool to quantify short-term driving styles.A short-term driving style evaluation method based on driving proneness was formed.The specific research content is as follows:(1)Analysis of factors affecting driving style and selection of parametersThe driving information parameters were classified from the four major aspects of human-vehicle-road-environment.The specific parameter set that affects driving style had been sorted out.Base on the idea that there is different driving proneness in different scenarios,the operation parameters of typical scenarios(car-following,lane-changing,overtaking)were screened.Finally,a parameter set of acceleration,braking,and steering operation used to characterize the driving style was obtained.(2)Research on the distribution characteristics of driving behavior data in typical scenariosReal road tests were carried out,which collected natural driving data of 10 drivers.Data in car-following,overtaking and lane-changing scenario were extracted.The time series symbolization algorithm was applied to reduce dimensionality and characterize operation parameter data.The distribution of operating characters was calculated to obtain the frequency distribution line chart of each driver.The fitted probability distribution curve was obtained through Gaussian and Gamma fitting.The differences in driving proneness of different scenarios were analyzed by comparing the fitting curves of different scenarios.The rationality of evaluating driving style based on scenario division was proved by the comparative conclusion.(3)Research on quantification method of short-term driving style based on driving pronenessEach three adjacent characters were combined to obtain driving operation fragments.The mapping relationship between the components of the fragments and the operation degree was established.The HMM parameter matrix corresponding to the characters in fragments was obtained by further statistics.The original character sequence data was input to the Viterbi model.The proneness state identification sequence was output,which included introversion,neutral,and extroversion.The extroversion value was calculated according to the proportion of extrovert and introvert data,which was used to quantify and characterize the driving proneness corresponding to specific parameters in typical scenarios.According to the calculation results of extroversion values,the entropy weight method was used to determine the corresponding weight of each parameter in the same scenario.The weighted extroversion values in the scenario were calculated,which were used to evaluate the short-term driving style of the current scenario.(4)Warning application analysis based on short-term driving style quantificationAccording to the extroversion values and the weighted extroversion values,a complete qualitative and quantitative short-term driving style evaluation of 10 drivers was carried out.Each driver was assigned an operating characteristic label.The early warning information transmitted to surrounding vehicles was determined.In different scenarios,specific driver groups were selected for early warning application case analysis.The practical application of the method was also prospected.The short-term driving style evaluation method can quantitatively and qualitatively describe the driver’s acceleration,braking,and steering operation characteristics based on the short-term data of the in-vehicle bus and the driving proneness recognition.The purpose of short-term driving style evaluation is achieved.In addition,the method can also be used for personalized parameter adjustment of intelligent vehicle driving systems.