When motors and generators fail, invisible electrical discharges through insulating materials, so-called partial discharges, are often to blame. Over time, their effects can accumulate, causing the materials to age prematurely, potentially leading to a short circuit.
For his Master’s thesis at ABB in Baden, Switzerland, Abbasi Sureshjani Samaneh developed a fully online diagnostic tool to detect the source of partial discharge pulses and plan appropriate remedial action before the device fails.
His tool automatically tracks partial discharges and groups them into clusters, which, using a fuzzy algorithm, can be mapped to a single source. Precisely determining the source is crucial, because the risk associated to each partial discharge pulse depends primarily on its source, not on its magnitude or frequency.