基于径向基函数网络的一步超前预测控制研究  

Research on One-step Ahead Predictive Control Based on Radial Base Functions Network

作　　者：丁国锋[1] 王孙安[1] 林廷圻[1] 史维祥[1] 

机构地区：[1]西安交通大学机械工程学院

出　　处：《控制与决策》1996年第4期485-489,509,共6页Control and Decision

基　　金：国家自然科学基金

摘　　要：提出一种基于径向基函数（ＲＢＦ）神经网络的一步超前预测控制算法。该方法只用一个网络，控制量的获取只求几步迭代，算法简单并有较好的实时性。通过对离散非线性系统的仿真证明了算法的有效性．In this paper, a one-step ahead predictive control algorithm based on Radial Base Function (RBF)neural networks is proposed, which needs only one neural network. The algorithm is simple, and the control value can be acquired by only a few iterations. Thus it has good real-time property. Through the simulation for a discrete nonlinear system, it is obvious that the algorithm posseses good control performance such as robustness.

关 键 词：神经网络 预测控制鲁棒性 径向基函数 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]