A failure-mechanism-and-data integrated system degradation analysis

In this research, a failure-mechanism-and-data integrated system degradation analysis method is proposed, in contrast with the traditional statistical data driven methods which are limited by predetermined statistical hypotheses and sparse degradation data. The effects of various failure mechanisms on the degradation of system performances are modelled by the modeling and simulation technique, which is developed by the integrated design of system reliability and performance. To meet the computation challenge of the expensive simulation process and maintain the time-varying characteristic, the time-dependent generalized polynomial chaos is employed to develop a surrogate model, the coefficients of which are calculated by the moving least square approach. To combine the surrogate model and statistical degradation data, an iterative Bayesian integration method is proposed for both cases of the single degradation trajectory and the multiple degradation trajectories. A case of a piston pump with 2700 hour leakage data illustrates the efficiency of the proposed method. Moreover, the proposed method is able to derive the degradation model from a single degradation trajectory, which is not possible for the statistical methods. The proposed method is also suitable to arbitrary performance metrics.