Variational mode decomposition based symmetric single valued neutrosophic cross entropy measure for the identification of bearing defects in a centrifugal pump

Abstract

This work contributes towards the development of a new approach of automatic identification of defects present in the bearing of a centrifugal pump using symmetric single valued neutrosophic cross entropy (SVNCE) of Variational Mode Decomposition (VMD). VMD aims at decomposing of sound signals into sparse distinct modes which are dense around the central frequency and thus makes the decomposition almost free from mode mixing problem. After computing the energy of each VMD mode, energy eigen values for the unknown bearing defect conditions are extended to establish energy interval ranges and thereafter converted these ranges into the form of single valued neutrosophic sets (SVNSs). Using minimum argument principle, the minimum SVNCE values between SVNSs of testing samples (collected from unknown bearing defect conditions) and SVNSs of training samples (collected from known bearing defect conditions) is utilized to identify the bearing defects in the centrifugal pump under study. The proposed method is intelligent in comparison with the existing methods because the overall accuracy retuned by our method based on SVNCE of SVNSs is 100% which is much better than retuned by the existing methods based on improved cosine similarity measure and fuzzy cross entropy measure of vague sets having relative accuracy of 76.4%.