Research on Transformer Failure Prediction and Fault Diagnosis Based on GM-BP and PNN

摘要：

　　This paper mainly researches transformer fault diagnosis and forecast based on dissolved gases in oil, and selects hydrogen, methane and acetylene, ethane, ethylene as the main characteristics of gas.First of all, the trend value of gases can be predicted by the model which includes grey theory model and BP neural network model. Taking the gray theory model output value as attribute value, BP neural network model can work out secondary predicted value and differences.Calculating these two values, we can finally get the predicted value of the gases.Second,multi-level probabilistic neural network for fault diagnosis model can analyze fault by using predicted value of the gases.Then,combining with the failure diagnosis and electrical properties of test data, this model will analyze which part of transformer has fault.

关键词： Artificial neural network Dissolved gases in oil GM-BP Gas forecast PNN Transformer Fault diagnosis

作者: Yaoye Zhu

作者单位: Shanghai Pudong Power Company

会议类型: 国际会议

会议名称: 第六届现代电力系统自动化和保护国际会议

会议地点: 南京

会议语种:英文

页码: 890-893

在线出版日期: 2015-09-21（万方平台首次上网日期，不代表论文的发表时间）