Addressing the limitations of a single fault detection model,inadequate for detecting complex multimodal pro-cesses arising from distinct process characteristics and statistical feature differences within the condenser system during op-eration,this paper proposes a multimodal condenser fault diagnosis method.This method integrates the K-Means clustering algorithm with principal component analysis to construct a fault detection model and acquire fault state samples.Through the analysis and separation of fault variables using restructured principal component analysis,the method generates fault feature vectors by considering the reconstruction residuals of the fault variables,allowing a classifier to identify fault categories.
NETL
NSTL
万方数据
