报告人: 周世荣博士（温州大学）

报告时间：2023年11月18日，16:10-17:00

报告地点：金融工程研究中心105报告厅

报告摘要：Stochastic processes play a crucial role in degradation modeling within the field of reliability engineering. However, extracting lifetime information from product degradation observational data has long been plagued by inefficiencies in modeling techniques and statistical inference methods. For commonly used Wiener, Gamma, and Inverse Gaussian process degradation models in reliability engineering, we comprehensively employ various approximate inference methods within the variational Bayesian framework to address the interdependence and skewness issues in the posterior variation of model parameters, aiming to achieve precise inference of model parameters. Additionally, we proposed a parameterized Gamma process degradation model and provided its variational Bayesian inference scheme. Compared to traditional Gamma process degradation models, it offers a more intuitive physical interpretability. For comparative purposes, we present the statistical inference process of the proposed models under traditional estimation methods and validate the provided variational Bayesian algorithm through comprehensive numerical simulations and case studies.

个人简介：周世荣，博士毕业于华东师范大学统计学专业，现就职于温州大学数理学院统计与信息科学系. 研究方向为可靠性统计、贝叶斯统计，目前在IEEE Transactions on Reliability, Reliability Engineering & System Safety期刊发表论文4篇.

邀请人：徐礼柏，刘芳