Automated Machine Learning for Remaining Useful Life Predictions - End-to-end AutoML RUL Method

Aug 3, 2026

Upload by Ivy

Predicting the remaining useful life (RUL) of engineering systems is central to prognostics and health management, and data-driven methods have grown as they avoid detailed physical modeling. However, these methods often require specialized machine learning expertise. The AUTORUL approach automates end-to-end RUL prediction for domain experts by automatically building pipelines and combining fine-tuned standard regression models into a powerful ensemble. Evaluation on eight real-world and synthetic datasets shows a viable alternative to hand-crafted RUL models while reducing the need for ML skills.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Predictive maintenance and PHM

RUL prognosis approaches

Related Work

Remaining Useful Life Prognosis

Automated Machine Learning for Remaining Useful Life Predictions arXiv :2306 . 12215v2 [ cs .LG] 17 Jan 2025 Marc-André Zöller§ USU Software AG Karlsruhe, Germany [marc.zoeller@usu.com](marc.zoeller@usu.com) Marius Lindauer Institute of Artificial Intelligence Leibniz University Hannover Hannover, Germany [m.lindauer@ai.uni-hannover.de](m.lindauer@ai.uni-hannover.de) Fabian Mauthe§ Esslingen University of Applied Sciences Esslingen, Germany [fabian.mauthe@hs-esslingen.de](fabian.mauthe@hs-esslingen.de) Peter Zeiler Esslingen University of Applied Sciences Esslingen, Germany [peter.zeiler@hs-esslingen.de](peter.zeiler@hs-esslingen.de) Marco F. Huber Institute of Industrial Manufacturing and Management IFF, University of Stuttgart Department Cyber Cognitive Intelligent (CCI), Fraunhofer IPA Stuttgart, Germany [marco.huber@ieee.org](marco.huber@ieee.org) Abstract—Being able to predict the remaining useful life (RUL) of an engineering system is an important task in prognostics and health management. Recently, data-driven approaches to RUL predictions are becoming prevalent over model-based approaches since no underlying physical knowledge of the engineering system is required. Yet, this just replaces required expertise of the underlying physics with machine learning (ML) expertise, which is often also not available. Automated machine learning (AutoML) promises to build end-to-end ML pipelines automatically enabling domain experts without ML expertise to create their own models. This paper introduces AUTORUL, an AutoML-driven end-to-end approach for automatic RUL predictions. AUTORUL combines fine-tuned standard regression methods to an ensemble with high predictive power. By evaluating the proposed method on eight real-world and synthetic datasets against state-of-the-art hand-crafted models, we show that AutoML provides a viable alternative to hand-crafted data-driven RUL predictions. Consequently, creating RUL predictions can be made more accessible for domain experts using AutoML by eliminating ML expertise from data-driven model construction. Index Terms—Remaining Useful Life, Automated Machine Learning, data-driven, RUL, AutoML, PHM, ML I. INTRODUCTION In recent manufacturing, a reliable, available, and sustainable production of goods is important to be competitive. This requires advanced maintenance strategies such as predictive maintenance. Traditional strategies such as corrective or preventive maintenance cause unplanned downtime or do not utilize existing resources completely due to superfluous maintenance actions. Predictive maintenance can help to avoid unplanned downtime while also reducing unnecessary maintenance costs. Knowledge about the future degradation behavior of an engineering system (ES) is crucial to plan the required maintenance as predictive maintenance is becoming more important in industry. The engineering discipline of prognostics and health management (PHM) studies techniques for transitioning from corrective or preventive maintenance to § These authors contributed equally predictive maintenance. A key task in PHM is the prognosis of the remaining useful life (RUL) . Approaches are often divided into model-based, data-driven, and hybrid methods [1] . Usually, the RUL is used to plan the next maintenance and has attracted considerable interest in the research community. Model-based approaches use mathematical descriptions, like algebraic and differential equations or physics-based models to predict the future degradation behavior of a system [2] . Such models require thorough understanding of the mechanism involved in the degradation process and are time-consuming to create. Also, not all systems can be expressed with sufficient precision in solvable mathematical or physics models due to the complexity of real world systems [3] . With the integration of more sensors, the amount of available data about the condition of a system is gradually increasing. In parallel,

FAQs

What problem does the document address in prognostics and health management?

It focuses on predicting the remaining useful life (RUL) of engineering systems to support predictive maintenance decisions.

Why are traditional data-driven RUL models not widely adopted?

They require specialized expertise in data modeling and machine learning, which many enterprises lack.

How does AUTORUL enable RUL prediction without ML expertise?

AUTORUL provides an end-to-end AutoML pipeline that automatically builds models and uses an ensemble of fine-tuned standard regression methods.

Aug 3, 2026

Upload by Ivy