A new adaptive technique for on-line partial discharge monitoring

The insulation system of HV apparatus deteriorates due to partial discharges (PD). PD tests are carried out successfully in shielded laboratories to verify the insulation quality and to rectify defects left during the manufacturing process. A variety of techniques exists for this purpose. It is also desired to monitor the insulation when equipment is in operation. Unfortunately, the performance of PD test equipment is adversely affected by interference signals. The sinusoidal signals of low and high frequency, due to their correlation, make the measurement difficult and sometimes even impossible. This paper describes the design of an adaptive digital filtering technique and its successful application to the on-line/on-site detection of PD current pulses generated in HV insulation in the presence of sinusoidal noise signals. The technique is based on Wiener optimal filtering theory and its characteristics are investigated. The adaptive filter of this system has a time-varying response and is self-tuning to each low and high frequency noise component. The filter remains active until the signal-to-noise ratio is significantly increased and it can be used for weak PD pulse integration. Theoretical background of the technique and the results of laboratory and field tests are presented