动态RBF神经网络在非线性系统建模中的应用

在分析传统RBF神经网络算法的基础上,提出了改进的最临近聚类算法和隐含层节点的增加与修剪策略,从而增强了RBF神经网络的非线性系统建模能力。以锅炉燃烧过程为实例,通过从现场采集的数据建立神经网络模型,使用改进的算法建立系统的初始神经网络模型,再经过改进算法不断修正模型,取得了比较满意的效果。

Application of Dynamic RBF Neural Network in Modeling of Nonlinear System

Based on careful analysis of the traditional algorithm of RBF neural network , an improved Nearest Neighbor-clustering Algorithm is proposed as well as a structural learning scheme with growing and pruning functions so as to improve its capability of modeling of non-linear system. Taking boiler combustion process as an example, a RBF neural network is built up and trained by the proposed method with pleased results.