摘要
对一种非线性时变系统提出了基于神经网络的自适应逆控制方案。该方案中用两个动态神经网络分别作为模型辨识器和自适应逆控制器,详细推导了在线训练自适应逆控制器的BPTM(backpropagationthroughmodel)和RTRL(realtimerecursivelearning)算法。根据大幅面喷墨打印机的结构特点,建立了打印头车架系统的时变非线性动力学模型作为仿真对象,在Matlab/Simulink平台下进行了算法仿真验证。结果表明了该方案收敛快,能有效控制该时变非线性对象。
An adaptive inverse control scheme was proposed for a kind of time-varying nonlinear systems. In the scheme there are two neural networks as the model identifier and the adaptive inverse controller. The BPTM and R TRL algorithms for training the controller were investigated in detail. To verify the scheme, a time-varying nonlinear dynamic model as the simulation plant was constructed for the printer head carriage of a kind of large scale ink jet printers with the special structure characters. The scheme was simulated in Matlab/Simulink, and the results show that the controller converges very fast and the scheme is suitable for the time-varying nonlinear plant.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第3期760-763,共4页
Journal of System Simulation
基金
国家自然科学基金(60443007
50390060)
关键词
自适应逆控制
非线性
神经网络
打印头车架
仿真
adaptive inverse control
nonlinear
neural network
printer head carriage
simulation
