主元分析和重构贡献在变压器故障检测与诊断中的应用

Appling Principal Component Analysis and Reconstruction-based Contribution to Fault Detection and Diagnosis of Transformer

针对常规DGA诊断方法对故障数据不敏感和诊断能力不足等问题,提出了基于主元分析和重构贡献的变压器故障检测与诊断方法。该方法在建立PCA模型后,基于故障重构的思想,定义重构贡献统计量并应用于变压器故障检测,分析了重构贡献法的故障诊断性能并利用重构贡献法进行诊断。实例研究结果表明:该方法的检测正确率超过95%且对检测DGA含量小的故障有一定的效果,同时,也验证了重构贡献法故障诊断的有效性。

In our opinion,dissolved gas analysis（DGA） is not sensitive to fault data and insufficient for diagnosis.Therefore we propose a new method for diagnosing the faults of a transformer with the principal component analysis（PCA） and the reconstruction-based contribution（RBC）.After the PCA model is built,we use the idea of fault reconstruction to define the RBC statistical indices and apply them to the diagnosis of the faults of a transformer.We analyze the performance of the fault diagnosis method that uses the RBC and perform the fault diagnosis.The example study results show that our fault diagnosis method has the effective fault detection rate of 95% and is somewhat effective for detecting the fault whose DGA content is small and that the fault diagnosis method that uses the RBC is also effective for the diagnosis of faults of a transformer.