Bivariate Correlation Degradation Reliability Assessment of Rubber V‐Belts Using the Inverse Gaussian Process

Rubber V‐belts are critical for reliable power transmission, yet conventional lifetime‐based reliability assessment is often costly, time‐consuming, and insufficiently accurate. To address this, this study proposes a bivariate degradation modeling approach that significantly enhances prediction precision—achieving errors below 5%—as validated by experiments on B1499‐type V‐belts. The method first establishes univariate degradation models for tension force and slip rate using the inverse Gaussian process to capture individual product variability. It then integrates these processes via a Copula function to explicitly model their strong statistical correlation, which univariate methods inherently neglect. Model parameters are estimated through a two‐stage Bayesian framework employing Markov Chain Monte Carlo (MCMC) simulation. Experimental degradation data confirms that the joint modeling of correlated degradation mechanisms yields more accurate and realistic reliability estimates than traditional univariate approaches. This work provides a practical, data‐driven framework for improved reliability assessment and condition‐based maintenance planning in belt drive systems.