An envelope statistic features and MFO-SVM fusion method for tidal stream turbine blades impact fault diagnosis

Abstract

Abstract The fault diagnosis of a Tidal Stream Turbine (TST) blade impact fault benefits its stable operation. However, when the stream velocity changes, it is hard to distinguish between different fault severities directly by observing the changes in the signal feature. To address this problem, this paper proposes a blade impact fault diagnosis method based on envelope statistical features and MFO-SVM. The method is divided into two parts. In the first part, the Teager-Kaiser energy operator (TKEO) and sliding window technique are introduced to extract the envelope statistic features of the current signal, and then the local outlier factor (LOF) values of fault sample points are calculated to form a new set of feature samples; in the second part, the fault feature samples are input into the support vector machine (SVM) optimized by moth fame optimization (MFO) for fault diagnosis. The experimental results show that the proposed method is more accurate than traditional fault diagnosis methods.