Parameter Identification of Rolling Element Bearing System Using Particle Swarm Optimisation Algorithm: An Application to Fault Diagnostics

Abstract

Model-based diagnostic techniques have been successfully used in the past in analysing the health of rolling element bearing machines. One of the advantages of model-based approach is the ability to incoporate physical understanding of the system
monitoring. Moreover, if understanding of the system degradation improves, the model can be adapted to increase its accuracy
and to address subtle performance problems. However, a poor accuracy of the model of the fully assembled machine can affect the accuracy of fault identifications since it is very challenging to identify parameters of a complex model thereby accurately capturing the behaviour of a system. Furthermore, model-based diagnostics requires a model that is very specific to the component and its geometry.
This paper introduces an automatic parameter identification based on Particle Swarm Optimisation (PSO) algorithm to identify the dynamic parameters of a rolling element bearing, thereby reducing the time taken to construct a model. With parameter identification, the base model can be adapted to different bearing sizes and geometry. In this paper, the vibration model for rolling element bearing with fault is derived. Construction of this model is done in MATLAB/Simulink environment by use of assumed bearing parameters. Particle swarm optimisation is then employed to obtain the global optimal solution of the bearing parameters. These optimized bearing parameters are then used in diagnosing bearing faults. To evaluate the feasibility of the approach, two publicly available data sets were employed. The results showed an average accuracy of 99.67\% and 99.2\% for bearing faults of Case Western Reserve University and University of Paderborn datasets, respectively.