Diagnosis of solid insulation deterioration for power transformers with dissolved gas analysis-based time series correlation

Dissolved gas analysis (DGA) is a prevailing methodology being widely used to detect incipient faults in power transformers. Although various methods have been developed to interpret DGA results, they may sometimes fail to diagnose precisely, especially when solid insulation deterioration is involved. This study presents a time series correlation technique, in which the sampled data of dissolved gases in the transformer oil are considered as a time series and the series correlation scheme in statistics is adopted to explore and manipulate the fault information. Both the constant and variable characteristic parameters are initially chosen based on analysis in terms of frequency histograms. With quantitative correlation analysis between the constant and variable characteristic parameters, a new criterion for solid insulation diagnosis is thereby proposed, which can be used to diagnose whether solid insulation deterioration is involved in a transformer fault, as well as further distinguish the electrical faults from the thermal ones. According to the proposed technique, explorative tests regarding two power transformers have shown promising results. With regard to the 91 fault samples collected from China Power Grid, the diagnosis accuracies for electrical and thermal faults were 86.5 and 77.7%, respectively, whereas it was 61.5% for the prevailing CO₂/CO criterion.