Title：Health Status Assessment of Industrial Equipment: Challenges, Methods, and Applications

Abstract：In most real-world scenarios, industrial equipment operates under highly variable conditions, where data available from new operating environments are often limited, and data collected under different conditions typically violate the assumption of independent and identically distributed samples. Consequently, health status assessment of industrial equipment faces two key challenges: 1) How to efficiently learn key indicators from small samples at early degradation stages? 2) How to effectively handle distribution shifts across multiple operating conditions? This talk presents an AI-centric workflow that addresses these challenges in a device-agnostic way with a representative case study of lithium-ion battery health management. The workflow comprises two complementary components. First, to address the challenge of limited early-stage samples, an early information utilization strategy is introduced to transform sparse initial observations into remaining useful life prediction and health status estimation. Second, to cope with distribution discrepancies among multiple operating conditions, a domain adaptation approach is employed to learn transferable representations across individuals and conditions, thereby enhancing the dynamic adaptability of health status assessment model. Together, these elements provide a device-agnostic route to more reliable health status assessment under small samples and diverse operating conditions.

Speaker：Liu Weibo，Brunel University.

Date：4:00 p.m., 2025-11-20 (Thursday).

Venue: 扬州大学瘦西湖校区数学学院208报告厅

Organizer：School of Mathematics

Inviter：Liu Yurong

Students and teachers are welcome.