Classification of partial discharges aims at the recognition of discharges of unknown origin. This classification is vital for the evaluation of discharges in tested constructions. For a long time, classification was performed by eye, studying discharge patterns at the well-known ellipse at an oscilloscope screen. In later years the introduction of digital processing techniques allowed automation of the recognition procedure. This paper reports on these procedures and applies them to a number of actual HV constructions which suffered from partial discharges.The results of these tests showed that a quite satisfactory recognition of discharges took place.

Making use of a computer-aided discharge analyzer, a combination of statistical and discharge parameters was studied to discriminate between different discharge sources. Tests on samples with different discharge sources revealed that several parameters are characteristic for different types of discharges and offer good discrimination between different defects.

No abstract

The time dependence of the discharge processes in dielectric bounded voids of 40 mm have been studied under power frequency voltage. Fast oscilloscopic techniques were used to observe the pulse shape of single discharges or clusters of discharges. Synchronous optical registration of discharge images was performed using a highly sensitive video system. At least two distinct discharge types were registered as a function of time. A 'streamerlike' discharge regime is gradually replaced by a more diffuse 'Townsendlike' discharge regime. This transition could be recognized in the shape of the discharge pulses and in the pattern of the optical image. A key parameter governing the transition is the surface resistance of the void-dielectric interfaces. This surface resistance was seen to decrease gradually due to discharge by-products. In its turn, a low surface resistance strongly favours the transition from 'streamerlike' to 'Townsendlike' discharges.

No abstract