A bivariate inverse Gaussian degradation process induced by a common random effect with RUL prediction for wet clutches

With the growing complexity of integrated systems, accurate degradation characterization now requires monitoring two or more performance characteristics (PCs). It is of great significance to effectively utilize the PCs for degradation analysis and online remaining useful life (RUL) prediction. Motivated by this, a flexible bivariate inverse Gaussian process model that incorporates degradation state-related measurement errors is developed. The common latent variable is introduced to capture cross-PC dependency. Then, the Bayesian method is employed to perform parameter estimation across varying sample sizes, and the model's effectiveness is verified through the simulation study. The reliability inference and the dynamic time window method for RUL prediction are developed using Monte Carlo simulation. Finally, a case study is performed on different wet clutches to evaluate their key PCs, including friction coefficient and engagement time. The results confirm the effectiveness of the proposed model, showing a statistically significant improvement over conventional stochastic models.