RBF神经网络在电路故障诊断中的应用研究

Research on the Fault Diagnosis of Circuit Based on RBF Neural Network

电路系统在工业控制中起着极其重要的应用,随着电路越来越复杂,电路各节点间的关系呈现非线性关系,若节点发生故障,如何确定故障发生在何处成为一大难题。利用RBF(radical basis function)神经网络可以快速逼近任意非线性函数及良好分类能力的特点,来实现对电路系统的故障分类。通过RBF与BP方法的比较及实例分析可得出结论,RBF可以很精确地确定电路中的故障来源,在对电路故障诊断能力方面具有较多的优越性。

Circuit systems are important components of industry control system. These circuit behaviors are nonlinear, how to check out some node which is faulted becomes one hard work. Radial basis function has the capacity to quickly close any nonlinear function, and has the ability of classification. It is advantageous for circuit system diagnosis. One example is test by RBF network and BP network, and the result saw the advantage of RBF network.