Abstract: The nonlinear dynamic model of automotive semi-active suspension is established with considering of the nonlinear characteristics of the MR (Magnetorhelogical) damper and the nonlinear rigidity of springiness element. At first, the non-linear control strategy of differential geometry theory is applied to execute feedback control on the semi-active suspension. The nonlinear model of the semi-active suspension is transferred to a simple linear system through a nonlinear state feedback. Then, according to the road excitation, the predetermined control object and the fuzzy control strategy to adjust the fuzzy logic control parameters, the fuzzy controller is designed. Furthermore, the neural network controllers are designed to improve automotive ride comfort. Finally, the simulation results are made a comparison between three non-linear control methods. It is shown that the suspension used fuzzy logic and neural network control methods has less impact response and lower vibration intensity than differential geometry theory control strategy, and has more superior performance.

Key words: Differential geometry, Fuzzy logic, MR damper, Neural network, Non-linear control, Semi-active suspension