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了这一方法是非常有效的． By constraining the system, a constrained multivariable generalized predictive control problem is constructed and T-HNN is used to solve this complex optimization problem. In the process of solving, the solution of constrained generalized predictive control is transformed into a steady-state solution of a T-H optimization network. So you can get through the hardware circuit or Runge-Kutta numerical method. The simulation study on an industrial process model proves that this method is very effective.