基于最小二乘加权融合集成神经网络的电力变压器故障识别_吕干云.pdfVIP

第28 卷 第 16 期 电 网 技 术 Vol. 28 No. 16 2004 年8 月 Power System Technology Aug. 2004 文章编号：1000-3673 （2004 ）16-0052-04 中图分类号：TM407 文献标识码：A 学科代码：470·4034 基于最小二乘加权融合集成神经网络的 电力变压器故障识别 吕干云，董立新，程浩忠 （上海交通大学电气工程系，上海 200030 ） POWER TRANSFORMER FAULT DIAGNOSIS BASED ON COMBINING NEURAL NETWORK WITH LEAST SQUARE WEIGHTED FUSION ALGORITHM LÜ Gan-yun，DONG Li-xin，CHENG Hao-zhong （Department of Electrical Engineering ，Shanghai Jiaotong University，Shanghai 200030，China ） ABSTRACT ：A new method to diagnose power transformer 关键词：变压器故障识别；集成神经网络； 最小二乘加权 faults based on the combination of neural network with least 融合；电力系统 square weighted fusion algorithm is presented. Firstly, the 1 引言 numerical preprocessing to the contents of five characteristic gases obtained by chromatography is performed and the six 电力变压器是变电站的枢纽设备，及时发现 characteristic quantities which are necessary to fault diagnosis 变压器内部故障的性质和发展趋势，是合理制定 are abstracted, and five back-propagation(BP) artificial neural 其维护措施的重要依据。油色谱分析是一种检测 networks(ANN) with different structures are applied to identify [1,2] 变压器内部故障性质的常用方法 。人工神经网 respectively, then the information fusion of the identified 络（ANN ）也逐渐用于识别变压器故障，目前已 results from the subnetworks is carried out by use of least [3-7] 做了很多研究工作 ，其中文[4]、[5] 的人工神经 square weighted fusion algorithm. Finally, according to the situation o