Stator interturn short circuit fault identification in DFIG and its analysis using an artificial neural network

Abstract

Inter-turn short circuit (ITSC) issues are a common electrical failure mainly caused by the deterioration of winding insulation in the machine over time. Failing to detect such issues early can lead to catastrophic consequences. This article investigates the interturn fault in the stator winding of a doubly-fed induction generator (DFIG) used in wind turbines. A flux linkage difference vector (FLDV) model is introduced in this study for fault detection. Additionally, an artificial neural network (ANN) model is proposed to classify these faults. Specifically, a short-circuit fault is induced in each phase of the stator winding, and the faults are classified by assigning different magnitudes to the respective phases. The ANN is trained to identify which phase contains an interturn fault, with output waveform amplitudes of 1, 2, or 3 corresponding to faults in phases "a," "b," and "c." If no fault is present, the waveform magnitude is designated as "0." This approach enables early fault diagnosis by analyzing waveform patterns, thereby preventing overheating caused by short circuits and avoiding severe, irreversible damage to the windings.