Application of Sparse Representation in SCADA: Wind Turbine Fault Diagnosis Based on K-SVD

Abstract

Wind turbine (WT) fault diagnosis is of great significance to maintain the high reliability and long-term safe operation cause the number of WT continues to grow. The most important component in WT is rotating machine, so the data monitoring and fault diagnosis of rotating machine is the top priority. The supervisory control and data acquisition (SCADA) system is usually used to monitor the rotating machinery in WT. However, hundreds of condition variables of SCADA make it a big challenge for diagnosis. Hence, the paper proposed a method combined data dimension reduction and sparse representation, which is implemented through spearman rank correlation and k-singular value decomposition (K-SVD) based on orthogonal matching pursuit (OMP). This method improves the efficiency of data processing through remove redundant data from a large amount of data. Then the dictionary and sparse coefficient matrix are obtained for the classification of test data. The results show that the method proposed in this paper can accurately classify test data.