摘要
通过对系统的信号约束,构成有约束广义预测控制问题.并实施一系列的转换步骤,将这一问题化为两个神经网络的求解平衡点问题.理论分析保证了这一求解是有约束预测控制问题的全局最优解.并设计了求解该问题的神经网络电路,使得有约束预测控制的求解能在电路的时间常数级内完成.
Through constraining to the signals of system,the constrained generalized predictive control method is constructed. The Problem is transformed into the solution of an equilibrium point of a neural network by performing a series of transformation steps. The theoritical analysis guarantees that the solution is the global optimum of the constrained generalized predictive controller. The neural network circuit is designed so that the solving process can be finished in the speed of the time-constant of the systems.
出处
《系统工程与电子技术》
EI
CSCD
1997年第7期61-63,66,共4页
Systems Engineering and Electronics
基金
天津市青年科学基金资助课题
关键词
神经网络
最优设计
广义预测控制
鲁棒算法
Neural network,Quadratic optimization, Predictive control, Constrain optimization, Global optimum.
