Wind Power Converter Fault Diagnosis Using Reduced Kernel PCA-Based BiLSTM

In this paper, we present a novel and effective fault detection diagnosis (FDD) method for wind energy converter (WEC) system with nominal power of 15 KW, which is designed to significantly reduce the complexity computation time possibly increase accuracy diagnosis. This strategy involves three significant steps: first, size reduction procedure applied training dataset, uses hierarchical K-means clustering Euclidean distance schemes; second, both reduced datasets are utilized by KPCA technique extract select most sensitive features; finally, in order distinguish between diverse WEC operating modes, selected features used train bidirectional long-short-term memory classifier (BiLSTM). study, various scenarios (short-circuit (SC) faults open-circuit (OC) faults) were injected, each scenario comprised different cases (simple, multiple, mixed on sides locations (generator-side grid-side converter) ensure comprehensive global evaluation. The obtained results show that proposed FDD via dataset methods not only improves but also provides an efficient storage space.