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。 Based on the analysis of the traditional RBF neural network algorithm, the improved nearest neighbor clustering algorithm and the hidden layer node adding and pruning strategy are proposed, which enhances the nonlinear system modeling ability of RBF neural network. Taking the boiler combustion process as an example, a neural network model was built by using the data collected from the field. An improved initial algorithm was used to establish the initial neural network model. The improved algorithm was used to continuously correct the model and satisfactory results were obtained.