变压器内部故障诊断研究  

Study on Fault Diagnosis of Transformers

作　　者：李霖[1] 刘飞霞[1] 杨章俊[1] 李娜[1] 贺莹[1] 

机构地区：[1]长沙理工大学,湖南长沙410003

出　　处：《电气开关》2013年第5期44-47,共4页Electric Switchgear

摘　　要：主要介绍以变压器油中溶解气体为分析对象,采用BP神经网络对变压器内部故障进行诊断。采用几种BP神经网络算法进行网络建模和仿真,并对它们进行了分析比较,得到了优于其他几种算法的将自适应学习率法和动量BP法相结合的学习率可变的动量BP算法,建立了一种具有较强学习能力、泛化能力和适应能力的神经网络模型,并通过实验证明了其对变压器故障诊断的准确性。This paper mainly describes that use BP neural network to analyse transformer oil dissolved gas and diagnosis transformer fault. In this paper, use several BP neural network algorithm to network modeling and simulation, analyses and compares these algorithm, get the variable learning rate momentum BP algorithm which is better than other algorithms that is the adaptive learning rate method combined with momentum BP method, estabilish a neural network model with strong learning ability and adaptability, the generalization ability, and experiment shows the accuracy of transformer fault diagno- sis.

关 键 词：神经网络 变压器 油中溶解气体 算法 

分 类 号：TM41[电气工程—电器]