基于BP网络模式识别的变压器故障诊断

Fault diagnosis of transformer based on BP network pattern identification

针对传统的离线色谱分析法在变压器故障诊断中的不足,提出了一种基于BP网络模式识别的变压器在线故障诊断方法。阐述了模式识别的原理,建立了诊断模型,给出训练精度,通过将BP网络诊断结果与传统三比值法诊断结果的对比,验证了该诊断方法的准确性与有效性。

method of fault diagnosis on-line based on BP networks pattern identification is presented in order to overcome the disadvantages of off-line chromatopile analysis for the transformer. Here we expound the principle of pattern identification, build the faults diagnosis model and give the training precision. Comparing the diagnosis result of BP net with the one of traditional three-ratio method, the accuracy and effectiveness of the diagnosis method is proved.