高新技术在变压器故障诊断中的应用  

Application of HI TECH in Transformer Fault Diagnosis

作　　者：董云云[1] 杨海利 

机构地区：[1]烟台南山学院,山东烟台265713 [2]东海热电有限公司,山东烟台265713

出　　处：《烟台南山学院学报》2016年第4期58-59,共2页JOURNAL OF YANTAI NANSHAN UNIVERSITY

摘　　要：变压器故障诊断中存在收敛速度慢、容易陷入局部极小值的问题，通过对基于Levenberg-Marquardt算法的BP神经网络进行深入分析，并最终应用于变压器故障诊断。该算法通过优化BP神经网络的搜索方向，加快了网络训练速度，提高了网络训练的精度。通过仿真证明能够有效地解决变压器的故障，为变压器故障诊断提供了一条新途径。Transformer fault diagnosis in the presence of convergence speed slow and easy to fall into the local minimum problem, through in-depth study of BP neural network based on Levenberg Marquardt algorithm, and the final application to fault diagnosis of transformer. The algorithm speeds up the searching direction of BP neural network, and speeds up the training speed of the network, and improves the accuracy of network training.. Through simulation of the example data, it is proved that this method can diagnose the transformer fault effectively, and provide a new way for the transformer fault diagnosis.

关 键 词：电力变压器 L-M算法 BP网络 故障诊断 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]