Particle swarm optimization-based kurtosis maximization in fractional Hilbert transform for bearing fault diagnosis

Abstract

This paper explores the application of Fractional Hilbert Transform (FrHT) for bearing fault diagnosis. Though recent publications portray the use of FrHT and show that kurtosis varies significantly with the fractional parameter (p), no attempt is made to calculate the optimum value of p. We propose the use of particle swarm optimization to maximize the kurtosis and get a better performing FrHT for bearing fault diagnosis. Simulation results show improvement of kurtosis over traditional Hilbert transform. Bearing fault sizes of 7 mil, 14 mil and 21 mil are more separable using the proposed method than the conventional Hilbert transform.