An ECT-LSLDA based fault diagnosis method for marine current turbines under variable marine conditions

Abstract

The development of marine current energy productions has recently gained attention. However, the variation in marine current will lead to weak fault feature and fault misclassification of marine current turbines (MCTs). This paper proposes a new method called extended Concordia transform-latitudinal summation linear discriminant analysis (ECT-LSLDA) to alleviate this problem. First, the extended Concordia transform (ECT) method projects the acquired three-phase currents into the stationary coordinate system to achieve the derivative modulus signals as an effective fault feature. Then, the latitudinal summation linear discriminant analysis (LSLDA) method normalizes and classifies the modulus to realize MCT faults recognition. Experimental results show that the proposed fault diagnosis method achieves high accuracy compare with the state of the art.