A two‐stage transformer fault diagnosis method based multi‐filter interactive feature selection integrated adaptive sparrow algorithm optimised support vector machine

Abstract The scarcity of samples and disunity feature inputs hinder the enhancement transformer fault diagnosis performance, there are mutual influences between model construction selection, which cannot only consider a single process. Therefore, this study proposes novel two‐stage strategy, includes multi‐filter interactive selection method (MIFS) constructed, ASSA‐SVM based on adaptive sparrow algorithm (ASSA) optimised support vector machine (SVM). Firstly, proposed MIFS incorporates ReliefF mRMR to establish comprehensive criterion ReliefF‐mRMR for importance ranking, then performs dimension‐by‐dimension input classifier interaction ranking results obtain optimal subset. Secondly, ASSA was optimise kernel parameters SVM. A integration MIFS‐ASSA‐SVM developed. Finally, Experiments were conducted using real data, diagnostic performance different inputs, optimisation algorithms classifiers compared. show that well parameter optimisation, can dynamically interactively select subsets with few dimensions good generalisation its overall accuracy reached 92.47%, each type has in multiple evaluation metrics.