Research On The Key Technology Of Driver Kinematics Simulation

With the increase of the vehicle quality,more and more attention have been paid for the requirement of driving comfort.In the design process of automobile products,a large number of ergonomics analyses must be carried out.However,most of the analyses are mainly based on statistical methods,which are time-consuming and laborious and are not easy to conduct.Therefore,the key techniques of driver ergonomic analysis are desirable to be studied,which are the main focus of this thesis.Based on various mathematical methods,the driver kinematics models were established,in order to be used in the modeling simulation of the driver human body.These models were validated for their feasibilities.Firstly,from one-dimensional and two-dimensional perspectives,this paper analyzed the differences of body size between five countries: the US,Canada,France,Japan and Korea.In one-dimensional body size,a differential analysis was conducted from individuals and groups.In the aspect of two-dimensional human body size,the confidence interval of the two-dimensional human body size of the five countries were established and analyzed.Analysis showed that there were indeed great differences between human dimensions in different countries and ethnicities.Secondly,us100,us94 and us125 of human dimensions were selected as predictors.Based on the GS-RBF interpolation method,various human body size prediction models were established.The models were analyzed for errors and compared with the traditional regression model.The results showed that RBF interpolation and regression analysis had good prediction effect for predictors with better correlation,and RBF interpolation was better than regression analysis.When the correlation of predictors was poor,the use of RBF interpolation could greatly improve the prediction accuracy.Then,based on the driver's seated reach data provided by the UMTRI's biological science laboratory,the regression analysis,BP and RBF neural network,GS-RBF and MQ-RBF interpolation methods were used to establish the prediction models of the driver's extension posture angles and difficulty ratings,respectively.And these models were verified and compared separately.The verification results showed that the BP neural network and radial basis function interpolation method were superior to the regression analysis,and the traditional regression analysis had a better prediction effect than the RBF neural network.The prediction accuracy of neural network and RBF interpolation is related to the relevant parameters in their respective functions,and different types of sample data will correspond to different optimal parameter values.In addition,the body size(height S,sitting height H,BMI)and spherical coordinates of target points(?)were also selected as independent variables,by using GS-RBF interpolation method to establish the driver's seated reach capability interface of different difficulty ratings,and compared with the SAE standard.The results showed that the difficulty of SAE seated reach envelopes in the middle level.Finally,based on the CPM model,the eye ellipses of different percentiles were obtained by simulating the driver's main driving posture.And the relevant parameters of the 95 th and 99 th percentile eye ellipses were calculated and compared with the parameters of the SAE eye ellipse standard.It showed that the simulation result of the 99 th eye ellipse was superior to the 95 th eye ellipse.It also showed feasibility of the method presented in this paper.