Meta-Analysis of Data from Radiometric Partial Discharge Localization Systems

Partial discharge localization is a crucial step in monitoring high-voltage equipment. Over time, due to data acquisition limitations, signal strength has become a key driver of methods of localization. In the literature, the variety of testing conditions makes it challenging to identify the factors influencing localization accuracy clearly. Furthermore, there is a lack of research using real equipment as a source of partial discharge. This paper presents a meta-analysis of data from monitoring systems used for radiometric localization of partial discharge sources, employing algorithms based on the time difference in arrival and received signal strength indicator. The obtained results show that the data exhibited errors with a high standard deviation, indicating that many studies are conducted under significantly different conditions. The first method is more sensitive to noise and the sampling rate, while the second is influenced by the source type and the number of sensors. Moreover, laboratory experiments performed with a potential transformer exhibiting partial discharge suggest that signal propagation in real equipment might not be omnidirectional, which can affect accuracy. Consequently, results based on controlled sources, where source directionality and obstacles have a significant impact on performance, do not accurately reflect real-world conditions. The obtained results indicate that the more realistic the source, the higher the errors will be, due to the lower uniformity in signal propagation.