Failure analysis of a lock mechanism with multiple dependent components is studied. Wear of the lubricated revolute joints in the mechanism presents the characteristic of multi-phase due to the change of degradation mechanism. It is not suitable to govern the degradation by stochastic processes only considering single phase pattern. In this paper, a change-point Wiener and Inverse Gaussian process is modeled to capture the two-phase degradation of the revolute joints. And the degradation parameter estimation based on Bayesian method is addressed. Since the wear evolution of the joints in the mechanism is affected by the same influence random variables, there exists dependence between them, and Vine Copula is utilized to describe the dependence. Besides, the time-varying dependence between the degradation processes is also investigated. Then, the reliability of the lock mechanism is obtained, and the result shows that the time-varying dependent model can more accurately describe the reliability curve of the system. Finally, in order to quantify the contribution of each degrading component on the system failure probability, the component importance measure is extended dynamically, through which the revolute joints can be ranked so that the weakest joint will be identified afterwards.
Read full abstract