关键词: 主动悬架, 自适应控制, 神经元, 综合误差, 多变量

Abstract:

This paper investigated the adaptive control of vehicle active suspension, in which multivariable and uncertainty problems were involved. A novel multivariable neuron controller was proposed. An integrated error concept was firstly given, then a new neuron controller combining integrated error with traditional neuron controller was formed. It can directly control the multi-output variables of controlled plant simultaneously. The control scheme was applied for controlling a quarter-car active suspension, and the optimum parameters of the corresponding controller were obtained by means of minimization of the quadratic performance index. The self-adaptability of the controller was investigated in various vehicle parameters and random road conditions. Extension comparisons with an equivalent passive suspension and an active suspension using traditional neuron control were also performed. Simulation results demonstrate the proposed controller is a simple, effective and robust adaptive controller which can achieve better vehicle overall performance, especially in ride comfort. In addition, the newly-proposed neuron scheme provides a concise yet efficient new possibility to the control of a class of uncertain multivariable systems similar to vehicle active suspension.

Key words: active suspension, adaptive control, neuron, integrated error, multivariable