Wavelet Packet Decomposition Based Detection and Classification of Stator Winding Insulation Degradation for Electric Machines

For electric motors, monitoring the condition of winding insulation becomes necessary to ensure a safe and reliable operation, which can help in the prevention of faults like short-circuit. In this paper, a novel insulation condition monitoring technique has been proposed, which employs wavelet packet decomposition (WPD) to analyze high frequency (HF) line current and extract indicators for monitoring the state of health (SOH). Compared with existing methods, the proposed technique can provide the SOH indicators of turn-to-turn (TT) and groundwall (GW) insulation simultaneously through the analysis of line current. Moreover, antiresonance oscillations in the HF line current are also explored to determine insulation SOH, which contributes to a unique capability in classifying types of degradation. The procedures for insulation degradation detection and classification are summarized in a flowchart for simple implementation. To validate the proposed method, extensive simulation and experimental tests have been conducted. The proposed method demonstrates robust performance, ability to detect even a small amount of degradation and possesses a unique capability to classify and quantify different types of degradation.