A Novel Framework for Centrifugal Pump Fault Diagnosis by Selecting Fault Characteristic Coefficients of Walsh Transform and Cosine Linear Discriminant Analysis

In this paper, we propose a three-stage lightweight framework for centrifugal pump fault diagnosis. First, the vibration signatures are fast transformed using Walsh transform, and spectra obtained. To overcome hefty noise produced by macro-structural vibration, proposed method selects characteristic coefficients of spectrum. second stage, statistical features in time spectrum domain extracted from selected transform. These raw result hybrid high-dimensional space. Not all these help illustrate condition pump. issue, novel cosine linear discriminant analysis is introduced third stage. Cosine dimensionality reduction technique which similar interclass adds them to illustrative feature pool, contains key that represent achieve maximum between-class separation, then applied pool. This combination pool creation forms application analysis. The reduced set obtained given as an input K-nearest neighbor classifier classification. classification results outperform previously presented state-of-the-art methods terms accuracy.