补偿递归模糊神经网络及在热工建模中的应用

在传统的模糊神经网络中引入递归环节和补偿环节,构成了一种新型补偿递归模糊神经网络(CRFNN),改善了网络的动态响应特性和学习能力.在此基础上,采用一种新型序贯监督策略对网络进行结构辨识,能够有效地确定模糊规则的条数以及相关参数的初始值.针对CRFNN的结构特点,提出了改进的BP算法,能够对网络的结构参数进行进一步的学习.对典型的热工对象以及复杂的ALSTOM气化炉进行的建模计算结果表明,提出的CRFNN具有优良的动态响应特性和很强的学习能力,值得在热工建模与控制领域中推广应用.

Compensation-based recurrent fuzzy neural network and its application in modeling of thermodynamic objects

A novel compensatory-based recurrent fuzzy neural network(CRFNN) is proposed by adding recurrent element and compensatory element to the conventional fuzzy neural network in order to improve the dynamic response ability and learning ability.Based on this,a new sequential supervisory method is introduced for the structure identification of the CRFNN in order to effectively confirm the fuzzy rules and their correlative parameters.Furthermore,the back propagation(BP) algorithm is improved based on the characteristics of the proposed CRFNN to train the network.The results of modeling and prediction of the typical thermodynamic objects and the complicated ALSTOM gasifier show that the proposed CRFNN has excellent dynamic response and strong learning ability and it can be widely used in modeling and control of thermal process.