Fault Parameter Identification for Model Based Prognostics of a Ball Bearing With an Outer Race Defect Available to Purchase

This paper deals with identifying the fault parameters of a rotor-bearing system with an outer race defect. The fault parameters can then be used for prognostics. The faults in the bearing are modeled as pits in the outer race of a bearing in a rotorbearing system with four degrees of freedom. Discrete wavelet transforms are used to obtain the energy and entropy features of the rotor-bearing system. The relationship between the features and the fault parameter is studied. Particle swarm optimization is used to generate an optimal set of features. This optimal feature set is used to train an artificial neural network to determine the amount of fault in the system.