汽车半主动悬架的非线性神经网络自适应控制研究*

分析了汽车悬架的非线性特性，提出了基于神经网络的自适应控制策略，设计了神经辨识器和控制器，并通过一个补偿网络，来进行后悬架的预见控制。仿真计算表明，神经网络自适应控制的半主动悬架具有明显的减振效果，而加有后悬架预见控制其效果更佳。为验证仿真结果，还进行了台架试验。试验结果亦表明了半主动悬架的优良减振性能。

RESEARCH ON ADAPTIVE CONTROL FOR AUTOMOTIVE SEMI-ACTIVE SUSPENSIONS BASED ON NONLINEAR NEURAL NETWORKS

Based on the analysis of nonlinear characterishcs ofautomohve suspension, the suspension model is built, the adaptiveconbol stlategy based on neural network is Put forwald and theneed idennder and consuller are designed. In order to improve theconeal effects, a compensahng network is added to the controlsystem. It can be used for the Preview coned of the rear suspensionaccording to the road disotances experienced by the bout wheels.Simulation results show that adaptively consulled send-activesuspensions with nonlinear neural netWorks can achieve Obviousvibration reducing effects, and the effects are even better for thosewith Preview control of rear suspension. In order to check thesimulstion results, experiments are carried out on the test rig. TheresultS of different road surfaces and vehicle speeds coincide withthose of simulation quite well.