時 間：114年4月22日（星期二下午14:10~15:00）
地 點：管理學院新大樓M240
主講人：王義富 教授 (成功大學 統計學系)
講 題：Degradation Analysis of Multivariate Inverse Gaussian Process with Random Effects
Abstract
For high-reliability products, collecting sufficient failure data within a limited time is often challenging. To address this, degradation analysis is commonly employed to evaluate a product’s quality characteristic to estimate the product lifetime. However, with the advancements in technologies, analyzing a single quality characteristic for degradation is no longer sufficient. This makes the analysis of multiple quality characteristics necessary. In this study, we propose a method that incorporates a common dependent framework with the inverse Gaussian process to address multivariate quality characteristics. By utilizing a conjugate conditional random effect within the inverse Gaussian process, we develop the Multivariate Inverse Gaussian Process with Random Effects (MIGP) model. This model effectively captures both the heterogeneity among samples and the correlations among quality characteristics. Finally, simulation studies and a case study are conducted to validate the proposed multivariate degradation model.
Keywords:
Degradation analysis, multivariate quality characteristics, inverse Gaussian process, random effects