摘要
针对具有外部干扰等不确定因素的离散未知非线性受控对象,提出了一种鲁棒神经网络自适应控制策略。该策略运用自适应预测及带遗忘因子的递推最小二乘参数估计的思想,对神经网络的预报输出进行修正,利用鲁棒反馈控制器保证系统稳定性,并对控制信号的增量进行限幅以抑制突变大幅值干扰信号对系统的影响。将提出的控制方法应用于实验室级液面系统的仿真中,结果表明了该控制策略的有效性。
A robust adaptive control based on neural network for unknown nonlinear dynamical systems with bounded disturbances or unmodeled dynamics is proposed. It is realized using adaptive forecasting and the recursive forgetting factor least square method. The stability of system is guaranteed by a robust controller. The increment of control signal is restricted in a proper range, This scheme is applied to a laboratory-scale liquid-level system, and the results of simulation show the effectiveness of the proposed scheme.
出处
《控制工程》
CSCD
2007年第3期290-293,共4页
Control Engineering of China
基金
国家自然科学基金资助项目(60304012)
关键词
神经网络
模型参考自适应控制
自适应预测
液面系统
neural network
model reference adaptive control
adaptive forecasting
liquid-level system
