An acoustic sensor array approach for localising partial discharges in electric machines

The quest for reliable electric machinery is paramount, notably in the development of highpowered propulsion systems for electric vehicles, aircraft, and ships. With insulation defects implicated in approximately two-thirds of electrical machine failures, as indicated by recent studies, robust condition -based monitoring (CbM) strategies are essential for electrical failure modes. Among these, partial discharge (PD) signals the onset of insulation failure. This research presents an advanced approach for localising PD sources using acoustic sensor arrays in tandem with a generalised cross -correlation (GCC-PHAT) and time -difference of arrival (TDOA) algorithm. Analytical experiments are conducted on a traction -specific highpower switched reluctance motor geometry, validating the method's efficacy across diverse sensor array configurations. This pioneering analytical work informs the design of a dedicated acoustic source localisation system, encompassing a novel 32 -channel condenser microphone array. Through rigorous experimental validation, localisation accuracy with a remarkable subcentimetre precision in Euclidean distance using the GCC-PHAT and TDOA algorithms on sensor and embedded platforms was achieved. This precision is vital, facilitating the pinpointing and isolation of faulty coils, thereby forestalling catastrophic failures such as overheating and thermal runaway critical for enhancing CbM systems' efficiency and reliability.