A Semi-Supervised Approach with Monotonic Constraints for Improved Remaining Useful Life Estimation.

Diego Nieves Avendano,Dirk Deschrijver,Nathan Vandermoortele,Sofie Van Hoecke,Agusmian Partogi Ompusunggu,Colin Soete,Pieter Moens

doi:10.3390/s22041590

Diego Nieves Avendano, Dirk Deschrijver + Show 5 more

Open Access

https://doi.org/10.3390/s22041590

Copy DOI

Abstract

Remaining useful life is of great value in the industry and is a key component of Prognostics and Health Management (PHM) in the context of the Predictive Maintenance (PdM) strategy. Accurate estimation of the remaining useful life (RUL) is helpful for optimizing maintenance schedules, obtaining insights into the component degradation, and avoiding unexpected breakdowns. This paper presents a methodology for creating health index models with monotonicity in a semi-supervised approach. The health indexes are then used for enhancing remaining useful life estimation models. The methodology is evaluated on two bearing datasets. Results demonstrate the advantage of using the monotonic health index for obtaining insights into the bearing degradation and for remaining useful life estimation.

Highlights

Remaining Useful Life Estimation.Health assessment (HA) and remaining useful life (RUL) estimation of mechanical components is a key task in Prognostics and Health Management (PHM)
This paper presents a methodology for constructing PHM models for HA and RUL
The health index (HI) are informative in most cases, there are instances in which the offline HI may not be reliable

Summary

Introduction

Remaining Useful Life Estimation.Health assessment (HA) and remaining useful life (RUL) estimation of mechanical components is a key task in Prognostics and Health Management (PHM). Accurate PHM models allow improvements in terms of quality, safety, maintenance scheduling, and cost reduction. PHM techniques are grouped into three categories: model-based or data-driven and hybrid [1,2,3]. Model-based methods use analytical or physical models to approximate the component’s behavior and its degradation. Their main advantage is high accuracy and capability of simulating diverse scenarios, such as operating conditions or different component specifications, without having to run a physical experiment. The increased use of sensors in industries has led to an increased interest in data-driven techniques. Data-driven models use monitoring information to create models that approximate the component behavior and degradation

Methods

Results

Conclusion