Reliability assessment of the fault diagnosis methodologies for transformers and a new diagnostic scheme based on fault info integration

Dissolved gas analysis (DGA) is one of the most effective methodologies used for transformer fault diagnosis. However, the only adopted index, namely the accuracy rate, is not adequate for reliability assessment of the diagnosis schemes. A new approach to evaluate the reliability of the DGA-based transformer fault diagnosis methods is presented in this paper, in which both the accuracy rate and the sensitivity rate are chosen as the assessment indexes. With 628 transformer fault samples collected from all over China, a comparative reliability analysis is done with regards to seven prevailing diagnosis methods. The results show that, for the same fault type the accuracy rate and sensitivity rate do not always render consistence, and for different fault types the most effective diagnosis method is also different, which presents the basis for appropriate selection of the most suitable fault diagnosis method in practical applications and lay a foundation for further amelioration of the prevailing diagnosis schemes. In order to improve the diagnostic effectiveness and reliability, three new flexible diagnosis criteria for partial discharges, low energy discharges, mixture discharges and thermal faults are thereby proposed. Based on fault info integration, a multi-level and multiparameter diagnosis scheme is established, which possesses unique advantages for general applicability with preferable flexibility in the diagnosis criteria.