基于动态神经网络自适应控制的主动悬架系统

Adaptive Control of Active Suspension System Using Dynamic Neural-Network

主动悬挂系统中的液压执行器伺服系统是典型的仿射非线性系统,其参数也是时变的,要建立其准确的数学模型是很困难的。在该文中将主动悬架系统看作是一未知系统,利用动态神经网络对未知系统进行辨识,根据建模误差和参数的不确定性来调整网络的参数和结构并用动态神经网络的优化能力得出系统的控制率。在仿真试验中用动态神经网络所得到的控制算法对模型的输出进行跟踪控制,从仿真结果图可以看出这种控制器能获得较好的控制效果。

The model of hydraulic actuating servo system in active suspension system is difficult to set up, as it＇s a typical affine nonlinear system, at the same time, its parameters are varying with time. Active suspension system is treated as a black box system in this paper, dynamic neural network is used to discriminate the active suspension system at the same time molding error and are used to adjust the structure of network, The network＇s optimization ability is used to get the control parameter. Control parameter gotten by dynamic neural network is used to control the model ＇s output in simulation. The result concludes that the controller can get better control result.