Detection of oil-paper equilibrium moisture content in power transformers using hybrid intelligent interpretation of polarisation spectrums from recovery voltage measurements

Detection of moisture in oil and paper is a key factor in determining the health of insulation in a power transformer. High moisture content severely degrades the insulating strength of oil and paper and may eventually cause failure. It is therefore highly desirable to detect incipient failure. Recently, a method based on Recovery Voltage Measurement (RVM) on transformers has been successfully used by Pacific Power International in the Australian state of New South Wales to identify conditions such as contaminated bushings, traces of solvent in oil, effectiveness of vacuum dry out and treatment of oil, local moisture ingress, and presence of static charges in oil. A Matlab based hybrid Expert System-Neural Network software has been developed to interpret above conditions from measured RVM data on 60 (sixty) power transformers. In this paper, details of this non-destructive and non-intrusive technique and some interesting results are presented