A comparison study of hidden Markov model and particle filtering method: Application to fault diagnosis for gearbox

For gearbox fault diagnosis, it is expected that a desired fault diagnosis model should have good computation efficiency, and have good recognition ability in both fault detection domain and fault identification domain. Currently, there are mainly three type´s models in this area that are physical based model, artificial intelligence based model and data-driven based model. However, the first type model requires specific mechanistic knowledge and theory relevant to the monitored system structure which are hardly to realize; and the second type model needs large amounts of condition monitoring data which are also not always available; while data-driven model investigate proper statistical model to describe system state which is used widely in fault diagnosis domain. The purpose of this paper is to investigate two popular algorithms of date-driven models for gearbox fault diagnosis, namely hidden Markov model and particle filtering method. At the beginning, we briefly introduced the procedure of feature extraction and the theoretical background of this paper. Then we respectively proposed hidden markov model and particle filtering model for fault diagnosis. Finally, the comparison experiment was conducted for gearbox fault detection and the analysis results from this work showed that particle filtering method has better detection performance, while hidden markov model has better computation efficiency in this area.