Incremental knowledge-based partial discharge diagnosis in oil-filled power transformers

The abstraction of meaningful diagnostic information from raw condition monitoring data in domains where diagnostic expertise and knowledge is limited presents a significant research challenge. This paper proposes a means of abstracting the salient features required to characterize partial discharge (PD) activity detected in oil-filled power transformers. This enables ultra high frequency (UHF) sensor data to be interpreted and translated into a meaningful diagnostic explanation of the observed PD activity. Plant data captured from UHF sensors forms the inputs to a knowledge-based data interpretation system, supporting on-line plant condition assessment and insulation defect diagnosis. The paper describes the functionality of a knowledge-based decision support system, providing engineers with a comprehensive diagnostic explanation of partial discharge activity detected in oil-filled power transformers. The diagnostic output can then be used to advise the engineer in (and potentially automate) the classification and location of partial discharge defect sources