基于人工神经元网络模型的预测控制研究

ON THE STUDY OF PREDICTIVE CONTROL BASED ON ARTIFICIAL NEURAL NETWORKS

研究了基于人工神经元网络模型的非线性预测控制。所采用的网络为一种将线性模型与多层前向网络相结合的ＤＬＦ网络。仿真结果表明，该“混合网络”易训练，收敛速度可大大加快。在ＤＬＦ模型的基础上，本文研究了一种非线性预测控制算法，它的显著特点是在线计算量小。对于一非线性过程──球形罐液位的仿真结果表明，基于ＤＬＦ的非线性预测控制效果颇佳。

The study of nonlinear predictive control based on artificial neural networks is carried out. The neural network used is called DLF network, which is the combination of multilayer feedforward network with linear model. The simulation results indicate that the proposed 'Hybrid networks has a lot of advantages such as easy training, fast convergence and so on. Based on the DLF model, a nonlinear predictive control algorithm is proposed in the paper. The significant feature of the suggested control strategy lies in its low online computing burden. For a nonlinear process -- the level of a sphere tank, the simulation results have shown that the predictive control based on DLF neural network can obtain fairly good result.