基于Hopfield网络的控制有约束MAC算法在主汽压力系统中的应用研究

THE APPLICATION OF MAC WITH CONTROL CONSTRAINED BASED ON HOPFIELD NETWORK IN THE MAIN STEAM PRESSURE SYSTEM

模型算法控制(MAC)是预测控制的一种,在惯性迟延对象的控制中有很好的应用。但MAC算法存在稳态误差及控制作用约束性差等问题。Hopfield网络在求解有约束二次优化问题中有很好的应用。该文将MAC转化为Hopfield网络可以求解的形式,应用Hopfield网络来求解有约束的预测控制序列。通过对电厂主汽压力对象的仿真表明,由Hopfield网络求解的控制作用满足给定的约束条件并且具有很好的控制效果。

MAC has a good application in the control of some objects with big inertia and delay. But the MAC has some defects such as stable error and bad constraint of the control output. Hopfield network is good at solving the quadraticconstrained optimizing problem. In this paper, we convert MAC to the pattern of Hopfield network and use it to solve the control output series with constraints. The simulation of a main steam pressure object proves the good result.