基于PSO-RBF神经网络的模拟电路故障诊断研究  

Research on analog circuit fault diagnosis based on PSO-RBF neural networks

作　　者：郭珂[1] 伞冶[1] 朱亦[1] 

机构地区：[1]哈尔滨工业大学,黑龙江哈尔滨150001

出　　处：《电子设计工程》2011年第24期17-20,23,共5页Electronic Design Engineering

基　　金：国家自然科学基金资助项目(61074127)

摘　　要：针对模拟电路故障诊断的难点和传统诊断方法的不足之处,提出了一种基于PSO算法优化的RBF神经网络模拟电路故障诊断方法。为了约简网络结构从而提高诊断效率,采用主成分分析方法对故障特征进行有效提取。针对RBF网络传统训练算法中隐层节点中心及基函数宽度选取困难问题,提出采用PSO算法来优化训练RBF网络,以提高网络的训练速度和泛化性能。最后,通过电路仿真对所提方法的有效性进行了验证。According to the deficiency of traditional fauh diagnosis methods, the method based on particle swarm optimization and radial basis function neural networks is proposed. In order to reduce the input dimension of neural networks and improve fault diagnosis accuracy, the principal component analysis is adopted to abstract the fault characteristics. Because the width and center position of radial basis function for the hidden layer of RBF neural networks are difficult to select, the particle swarm optimization algorithm is adopted to train the RBF neural networks. At last, the efficiency of the proposed method is verified by the simulation circuit.

关 键 词：粒子群优化算法 径向基函数神经网络 主成分分析 故障诊断 模拟电路 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]