东南大学学报（自然科学版）Issue(4)：769-774,6.DOI:10.3969/j.issn.1001-0505.2014.04.016
基于差分进化与 RBF 神经网络的热工过程辨识
Thermal process identification based on differential evolution and RBF neural network
摘要
Abstract
For the nonlinear identification of thermal process, a new radial basis function neural net-work ( RBFNN) design method is proposed based on the differential evolution algorithm ( DE) .In the method, the population in the DE algorithm is divided into several parallel subpopulations, and each subpopulation corresponds to a class of RBF network solutions with the same hidden nodes.In the RBFNN learning process, the network structure and parameters are adjusted simultaneously through the parallel optimization of the subpopulations.Under the given error limit, the algorithm can design an RBF model automatically with fewer hidden nodes according to thermal input and out-put data.Then, the algorithm is used to identify nonlinear thermal processes.For single-input sin-gle-output system identification, only one node is required in the RBFNN hidden layer.For multi-in-put single-output system identification, the RBFNN model also requires less hidden nodes.The sim-ulation results show that the proposed approach can achieve the given identification accuracy with fe-wer hidden nodes, and has good generalization ability.关键词
热工过程/系统辨识/径向基函数/差分进化/建模Key words
thermal processes/system identification/radial basis function/differential evolution/modeling分类
信息技术与安全科学引用本文复制引用
薛晓岑,向文国,吕剑虹..基于差分进化与 RBF 神经网络的热工过程辨识[J].东南大学学报（自然科学版）,2014,(4):769-774,6.基金项目
国家高技术研究发展计划（863计划）资助项目（2006AA05A113-1）． （）
