Research On Fuzzy-Neural Network Semi-active Control Of New Magnetorheological Suspension

Semi-active suspension has become a hot issue in vehicle suspension research field,because of its good control performance,low energy consumption,low cost and so on.The control performance of vehicle MR suspension depends on the semi-active control strategy,whereas the conventional control methods have their own shortcomings.Therefore,a new intelligent control strategy based on fuzzy-neural network is proposed in this thesis.The quarter vehicle MR suspension is taken as the research object,and the semi-active controller based on fuzzy-neural is combined with fuzzy control and BP neural network algorithm(FNNC).In view of the traditional BP algorithm has slow learning rate and long iteration period,an adaptive learning method with faster iteration speed is proposed in the thesis,in addition,the fuzzy rules are matched with the hidden layer of the neural network.In the FNNC,the center and width of the fuzzy subset are learned.At the same time,the activation factor is introduced to weigh the iterative optimization of the fuzzy rules for achieving good control performance of the FNNC.In order to verify the control effect of the strategy on the semi-active suspension,the quarter vehicle MR suspension performance with FNNC is compared with the suspension performance of the modified Skyhook semi-active control under the three typical excitation signals(harmonic,smooth pulse and random road),according to the evaluation criteria of vehicle riding comfort and steering stability.The main performance indexes are quantitatively analyzed,involving damping force,driving current the sprung mass displacement,the sprung mass acceleration,the suspension relative displacement and the tire dynamic force.The results show that FNNC has better comprehensive performance.The proposed controller is further integrated with the quarter vehicle dynamic model,so as to systematically evaluate the controller robust and suspension performances under varying vehicle operation load,speed,as well as different road surface excitations.The results further illustrate that the proposed FNNC-based suspension system has good robustness on vehicle operating uncertainties,and can realize the ideal control requirements of vehicle ride comfort,driving safety and other multi-objective suspension performance.