Abstract: A new approach of Reconstruction-based Contribution(RBC) is proposed to solve the problems that traditional diagnosis method based on Dissolved Gas Analysis(DGA) is not sensitive to fault data. The SPE and T2 indices are used to fault detection after Principal Component Analysis(PCA) model is built up, the diagnosability of RBC method is analyzed and it is applied to the fault diagnosis. Case studies show that the proposed method is able to identify the reason of a fault effectively and enhance fault diagnosis accuracy, it has very good fault distinguishing ability and vast prospect.

Key words: Principal Component Analysis(PCA, Reconstruction-based Contribution(RBC), transformer, fault diagnosis

摘要： 在常规DGA诊断方法中，存在故障数据不敏感的问题。为此，提出一种基于重构贡献的变压器故障诊断方法。该方法在建立主元分析模型后，采用SPE和T2统计量检测故障，在分析重构贡献法故障诊断性能的基础上，利用重构贡献法识别故障。实例研究结果表明，该方法可以识别故障发生的原因，提高故障诊断的准确性，具有较好的故障识别能力。

关键词: 主元分析, 重构贡献, 变压器, 故障诊断

CLC Number: 

• TP391