Abstract
Most of the existing fault detection methods rarely consider the cost-optimal maintenance policy. A novel multivariate Bayesian control approach is proposed, which enables the implementation of early fault detection for a helicopter gearbox with cost minimization maintenance policy under varying load. A continuous time hidden semi-Markov model (HSMM) is employed to describe the stochastic relationship between the unobservable states and observable observations of the gear system. Explicit expressions for the remaining useful life prediction are derived using HSMM. Considering the maintenance cost in fault detection, the multivariate Bayesian control scheme based on HSMM is developed; the objective is to minimize the long-run expected average cost per unit time. An effective computational algorithm in the semi-Markov decision process (SMDP) framework is designed to obtain the optimal control limit. A comparison with the multivariate Bayesian control chart based on hidden Markov model (HMM) and the traditional age-based replacement policy is given, which illustrates the effectiveness of the proposed approach.
Highlights
In a helicopter system, the mechanical drive system is the most efficient and compact device to transmit torque and change angular velocity
In conditionbased maintenance (CBM) and prognostics and health management (PHM), the vibration monitoring data obtained from the accelerometers carry a large amount of information, which were used in the fault detection of the gears for decades and were very helpful to prevent unnecessary machine halts [1, 2]
We have proposed a novel optimal Bayesian control scheme for the early fault detection of the partially observable helicopter gear system
Summary
The mechanical drive system is the most efficient and compact device to transmit torque and change angular velocity. To the best of our knowledge, this is the first paper applying HSMM with general Erlang distribution of sojourn time in two hidden states and optimal Bayesian control scheme to find the cost-effective strategy for early fault detection of the helicopter gearbox. With the condition monitoring information, the proposed approach can update the remaining useful life at each sampling epoch, and process early fault detection of the helicopter gear system and determine the optimal stopping time with minimum cost simultaneously.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
