基于Tank-Hopfield神经网络的有约束多变量广义预测控制器

通过对系统的信号约束，构成有约束多变量广义预测控制问题，并运用Ｔ－Ｈ优化神经网络来求解这一复杂的优化问题。在求解过程中，有约束广义预测控制的求解被转化为一个Ｔ－Ｈ优化电路网络的稳态解。因此可以通过硬件电路或龙格－库塔数值方法进行求取。在一个工业过程模型上的仿真研究证明这一方法是非常有效的。

The Constrained Multivariable Predictive Controller Based on Tank-Hopfield Nenral Network

Through the constrain of the signals of system,this paper drives the constrained multivariablegeneralized predictive control,and solves the complicate optimizing problem with T-H neural network. It istransformed into the solving of a stable state of a T-H optimizing neural networks for the constrained generalized predictive controller. Hence,the solution can be obtained through hardware electricity circuit or RungeKutta numerical algorithm. The simulation research on a process verifies that the method is very effective.