摘要
针对电液位置伺服系统存在的强非线性、控制增益未知和非匹配不确定性,通过引入神经网络和带饱和层的多滑模面,提出了一种多滑模神经网络控制方法.该方法运用神经网络的万能逼近特性和滑模控制优良的抗干扰特点,采用构造性方法设计控制器.运用光滑投影算法和积分李雅普诺夫技术,避免了参数漂移和控制器奇异问题.理论证明了系统跟踪误差收敛于任意设定的滑模面饱和层内.仿真实验表明了理论结果的有效性.
A multiple sliding mode neural network control scheme, which combines the technology of neural network and sliding mode surface with boundary layer, is proposed for an electro-hydraulic position servo system with strong nonlinear terms, unknown control gain and mismatched uncertainties. By employing the universal approximation property of neural network and the advantage of sliding mode control, and using constructive method, the controller is designed. The parameter drift and controller singular problem are avoided perfectly by applying the smooth projection method and the integral type Lyapunov function. It is proved that the tracking error converges to the arbitrarily prescribed boundary layer. Simulation results demonstrate the effectiveness of the control strategy.
出处
《控制与决策》
EI
CSCD
北大核心
2009年第2期221-225,共5页
Control and Decision
基金
国家自然科学基金项目(60504013)
重庆大学青年骨干教师创新基金项目(CDCX019)
关键词
电液位置伺服系统
滑模控制
神经网络
非线性系统
Electro-hydraulic position servo system
Sliding mode control
Neural network
Nonlinear system