EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2885. Condition-based Maintenance for a Degrading System under Dynamic Working Conditions

Invited abstract in session WB-39: Reliability Models, stream Stochastic Modelling.

Wednesday, 10:30-12:00
Room: 35 (building: 306)

Authors (first author is the speaker)

1. GUO SHI
Management Science, University of Strathclyde
2. Bin Liu
University of Strathclyde
3. Lesley Walls
Management Science, University of Strathclyde

Abstract

This study proposes a maintenance strategy for a degrading system under dynamic working conditions. Dynamic working conditions including environmental factors such as temperature, and humidity, alongside operating factors like production rate and operating speed, present challenges in accurately capturing the degradation process. In addition, the heterogeneity among observed components adds complexity to utilising the health monitoring data. To address these issues, we utilise Bayesian Linear regression and Bayesian Poisson regression to account for shock occurrence and magnitude. During system operation, regression parameters are updated upon the arrival of shocks at each decision epoch. We formulate the optimal maintenance problem as a Markov decision process, with decisions triggered by parameter updates converging towards underlying values. This paper not only establishes an analytically tractable degradation model incorporating component heterogeneity and dynamic working conditions but also theoretically explores the structure of optimal preventive maintenance thresholds. Additionally, to ease the computational burden due to the high state space, we introduce a heuristic algorithm that leverages the most likely distribution. Numerical experiments and a case study validate the effectiveness of the proposed maintenance policy, demonstrating its practical applicability and efficacy in real-world scenarios.

Keywords

Status: accepted


Back to the list of papers