Control Strategy Of Intelligent Vehicle Suspension System Based On The Perception Of The Road

With the development of vehicle technology,intelligent suspension systems could ensure ride comfort along with good road handing and safety in ground vehicles.A vehicle is always subjected to random excitation due to an irregular road profile.Intelligent suspension systems are designed to absorb the energy and mitigate uncomfortable vibrations due to the random excitation.As a result of this,to create adaptive suspension systems,semi-active suspension systems are the current focus of extensive research to improve vehicle ride comfort and road handing.Control of an automotive semi-active suspension with road information is a new direction of research.This paper focuses on the control of vehicle vertical dynamics.Based on the road surface estimation with multi-senor data fusion,the objective function with different performance criteria is used in order to fully enhance the overall performance of the vehicle suspension.At first,McPherson suspension nonlinear model and quarter-linear model are presented.The detailed structure of a continuously variable semi-active damper based on physical insights is introduced.Depending on MTS bench testing,force-velocity characteristics with different forms can be achieved.Then,the radial basis function(RBF)neural network is used to approximate the forward and inverse dynamic behaviors of CDC damper.Training and validating of the RBF network models are achieved by using the data generated from the test and the results shows the high precision of the approximation.Non-dimensionalised suspension parameters are presented for analyzing the behavior of semi-active vehicle suspensions by using skyhook,groundhook,and hybrid control policies,and compared with passive suspensions.The relationship between vibration isolation,suspension deflection,and road holding is studied,using three performance indices.The dimensionless parameters,such as the mass ratio,the stiffness ratio and the damping ratio,are adopted to illustrate the effects of the parameters on the response of the quarter-car.It can be provide better insight into how the RMS responses are influenced by the vehicle model parameters.To solve the problem related to identifying vehicle sprung mass,the new method combined Kalman filter and Recursive least square(RLS)with forgetting factors enables simultaneous estimation of the piecewise constant mass.The Kalman filter is firstly utilized to estimate the velocity of unsprung and sprung masses,and the estimated variables are then applied by RLS to identify the mass.To improve accuracy of Kalman filter,the process noise covariance matrix is chosen based on the different road level excitations.A novel method is proposed to classify road excitation levels based on multi-sensor data fusion.Measurable signals of suspension system,including unsprung mass acceleration,rattle space and the wheel acceleration,are firstly sampled as candidate signals,wavelet packet and empirical mode decomposition(EMD)is then performed to obtain four components of different frequency ranges.Subsequently,Max-relevance Min-redundancy method helps to select salient features with best trade-off between relevance and redundancy.With these selected features,a cascade classifier with probabilistic neural network(PNN)and random forest is applied to output the classified road class.Due to different requirements of ride comfort and handling stability in various road levels,two different suspension controller solutions for the nonlinear quarter-car model are proposed and compared in this thesis.In addition,Different parameters would affect the performance of the active suspension dramatically even if they can all guarantee a stable system.Thus,the particle swarm optimization(PSO)technique is used to optimize the parameters for the control law based on the overall performance indices.The result for both the adaptive neural-sliding mode control(SMCNN)and adaptive backstepping control with neural networks(ANN)have been experimentally confirmed on a semi-active quarter-car test rig,which has been designed and constructed by utilizing production vehicle components.