Application of gene expression programming (GEP) to investigate the health condition of direct-drive wind turbine

Abstract

In this paper, Gene Expression Programming (GEP) based a wind turbine healthy condition identification model is proposed using generator current signals. Proposed GEP approach is capable to achieve very high classification accuracy. This is the first attempt to design such type of classifier using GEP for health condition identification of wind turbine. The beauty of proposed approach is to analyze the faults accurately with less process time. Moreover, proposed approach can also perform the self optimization process as it uses the function of both GA and GP in combine manner. Raw data of permanent magnet synchronous generator (PMSG) stator current is preprocessed through empirical mode decomposition (EMD) method to develop Intrinsic Mode functions (IMFs). Classifier uses the decision tree to further prune these IMFs to most relevant input variables which serve as input to GEP fault classifier. We compare performance of proposed GEP classifier with other AI based classifiers such as ANN and SVM. Obtained results and performance comparison shows that our proposed GEP based classifier could serve as an important tool for wind turbine fault diagnosis.