An ECT-PCA-based Fault Detection Method for Winding Asymmetry of Marine Current Turbine Generator

Abstract

The traditional detection methods of motor winding asymmetry often analyze the zero-sequence component. However, due to the different types of motors, the collection methods are also different. The marine current turbine (MCT) has a complicated sealing method due to the harsh marine environment, and its working conditions are frequently changed by the influence of the marine current flow rate, which makes it challenging to extract the fault characteristics. This paper proposes a novel method, called ECT-PCA, to detect MCT generator winding asymmetry, which includes: acquiring the stator three-phase current and using the extended Concordia transform (ECT) to obtain the modulus signal; dividing the modulus signal into an equal-length sample, and performing Fourier transform to obtain the frequency domain amplitude; Then establishing a PCA fault detection model, finally uses T2 and SPE statistics to detect whether the winding asymmetry or not. An experimental platform based on the MCT prototype was built to verify the effectiveness of the proposed method.