Estimation Of Road Adhesion Coefficient For Commercial Vehicles

With the improvement of people's living standards and the rapid growth of transportation,commercial vehicles as an important carrier of national economic development,their sales are also increasing year by year.Because of its high quality,commercial vehicles are prone to cause serious casualties and property losses when they lose their stability,which has attracted much attention in recent years.Road adhesion coefficient is an important parameter in vehicle stability control system,which can directly and objectively reflect the material of road and the actual condition of road surface.Real-time and accurate acquisition of road adhesion coefficient information can improve the safety of vehicles in the process of driving,and can effectively avoid traffic accidents.Therefore,the study of road adhesion coefficient estimation for commercial vehicles is of great significance.At present,the identification methods of vehicle road adhesion coefficient mainly include direct measurement method and indirect estimation method.Among them,the direct measurement method mainly relies on a variety of acoustic,optical and electrical sensors.This method is simple and effective,but considering the high cost and durability of sensors,it is difficult to popularize in commercial vehicles on a large scale in practical applications.However,the indirect estimation method does not need additional sensors,and has the characteristics of good effect and strong maneuverability,which has attracted wide attention of scholars at home and abroad.In recent years,some achievements have been made in estimating road adhesion coefficient based on slip rate curve,state observer and vehicle dynamics.However,there are some problems in estimating road adhesion coefficient using slip rate curve,such as large error,poor adaptability to road surface,and inaccurate estimation of strong non-linear state of vehicle dynamics combined with state observer.In the process of estimating road adhesion coefficient,the accuracy and convergence time of estimating road adhesion coefficient need to be further improved.In this thesis,aiming at the accuracy of estimating road adhesion coefficient,according to the different vehicle handling states,the road adhesion coefficient under straight and steering conditions is estimated by using Multi-Slip curve and neural network methods respectively.The main work of this thesis is as follows:(1)Firstly,vehicle dynamics and wheel forces are analyzed.A seven-degree freedom vehicle model and a tire model based on magic formula are established under the environment of Matlab/Simulink,and a closed-loop system is formed.Under three typical road adhesion conditions,the model is compared with the vehicle model in CarSim,which verifies the reliability of the vehicle model and tire model.(2)Based on the Multi-Slip curve estimation method,10 kinds of road surface adhesion-longitudinal slip curve data under pure longitudinal slip by magic formula tire model are calibrated in this study,then estimates and solves the current vehicle longitudinal slip rate and road surface adhesion rate separately,and finally obtains the current road surface adhesion coefficient by matching the calibration curve.In the estimation of longitudinal slip rate,Kalman filter is used to reduce the noise of sensor data,and the accurate value of road adhesion rate is calculated by combining with vehicle longitudinal force sensor.The simulation results show that the vehicle can get accurate estimates under acceleration,driving and braking conditions,and the road with different adhesion coefficient has good adaptability.(3)The estimation method based on neural network is suitable for vehicle steering condition.The non-linear function relationship between five important parameters of vehicle stability control and road adhesion coefficient are demonstrated in this study.A method of estimating road adhesion coefficient by optimizing BP neural network with genetic algorithm is proposed.The genetic algorithm solves the problem that BP neural network is easy to fall into local minimum,and reduces the output error.Considering the inaccuracy of traditional extended Kalman filter in estimating the sideslip angle of the center of mass of commercial vehicles during full-load steering,the iterative extended Kalman-assisted particle filter algorithm is used to improve the estimation accuracy and provide data basis for the neural network.Finally,the structure of neural network and the parameters of genetic algorithm are set up to verify the accuracy of the road adhesion coefficient estimation under steering conditions by setting up a variety of simulation conditions.