A Comparative Analysis of Machine Learning Models in Prognostics and Prediction of Remaining Useful Life of Aircraft Turbofan Engines - Master’s thesis

Aug 5, 2026

Upload by Quinn

This master’s thesis investigates how machine learning models and artificial neural networks can predict the remaining useful life (RUL) of aircraft turbofan engines within a Prognostics Health Management (PHM) framework. It uses the NASA turbofan degradation dataset as a case study and starts by establishing PHM and RUL principles. The work then introduces AI, machine learning, and deep learning, formulates regression-based approaches, and builds multiple models for a comparative evaluation. Data visualization, feature selection, data treatment, and hyperparameter tuning are applied to improve performance and support more reliable predictive maintenance decisions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages73

Reading Time3 h 4 min

AccessFree to download

Preface

Acknowledgment

Executive Summary

Introduction

Background

Objectives

Approach

Outline

Prognostic and Remaining Useful Life

Prognostics health management

Prognostics

Remaining useful life

RUL as a function of CM

RUL as a function of RF

Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL)

Artificial intelligence

Machine learning vs deep learning

Machine learning paradigms

Supervised learning

Unsupervised learning

Semi-supervised learning

Master’s thesis NT NU Norwegian University of Science and Technology Faculty of Engineering Department of Mechan ica l and Industrial Engineering Amirhossein Vakili A Comparative Analysis of Machine Learning Models in Prognostics and Prediction of Remaining Useful Life of Aircraft Turbofan Engines Master’s thesis in Reliability, Availability, Maintainability and Safety (RAMS) Supervisor: Professor Yiliu Liu October 2023 Amirhossein Vakili A Comparative Analysis of Machine Learning Models in Prognostics and Prediction of Remaining Useful Life of Aircraft Turbofan Engines Master’s thesis in Reliability, Availability, Maintainability and Safety (RAMS) Supervisor: Professor Yiliu Liu October 2023 Norwegian University of Science and Technology Faculty of Engineering Department of Mechanical and Industrial Engineering i Preface This report is written during the fall semester of 2023 in Reliability, Availability, Maintainability, and Safety (RAMS) at the Department of Mechanical and Industrial Engineering (MTP) at NTNU. My main drive in selecting this topic stemmed from a deep-rooted interest in the concept of prognostics and the prediction of the Remaining Useful Life of components and systems. Pairing this with a curiosity in programming and machine learning made this an exciting journey. While the thesis is primarily targeted at those with a foundational grasp of programming and reliability theory, the methods and techniques discussed, especially those related to machine learning, are introduced in a way that should be approachable for readers unfamiliar with the topic. Trondheim, 2023-10-31 Amirhossein Vakili ii Acknowledgment First and foremost, I would like to express my sincere gratitude to Professor Yiliu Liu for his invaluable guidance throughout this thesis journey. His insights were crucial in shaping my research direction and in navigating the complexities of machine learning and prognostic techniques. I’m also deeply appreciative of my fellow RAMS student, Mahshid Aghamalizadeh. Her consistent support, sharing of knowledge, and enriching discussions played a significant role in the formulation and completion of this report. Amirhossein Vakili iii Executive Summary In the pursuit of enhancing the safety and reliability of complex industrial systems, Prognostics Health Management (PHM) technology, especially its components of fault diagnosis and prognosis, has found increasing relevance in various industries. Fundamental to PHM technology is the prediction of the Remaining Useful Life (RUL). An accurate estimation of RUL not only leads to accident prevention but also facilitates optimal equipment maintenance scheduling in complex systems. This strategic foresight can decrease maintenance costs and spearhead amore systematic approach to predictive maintenance. While RUL prediction techniques span from model-based to data-driven approaches, the latter, which establishes relationships between RUL and historical data through a learning-based model, is gaining attention due to its independence from prior knowledge. Building on this foundation, this master thesis delves into the application of Machine Learning (ML) models and Artificial Neural Networks (ANNs) to predict the remaining useful life of turbofan engines, using the turbofan degradation dataset by National Aeronautics and Space Administration (NASA) as a case study. The research starts with an introduction to the principles of prognostic health management and RUL, establishing the context and relevance of the study. Subsequent chapters explore the domains of Artificial Intelligence (AI), underscoring its intersections with machine learning and Deep Learning (DL). A comprehensive overview of machine learning paradigms, including supervised and unsupervised methodologies, is provided, setting the stage for a deep dive into various regression models. Given the dataset’s properties and the continuous, labeled nature of

FAQs

What problem does the thesis address?

It addresses predicting the remaining useful life (RUL) of aircraft turbofan engines to improve safety and reliability and enable better predictive maintenance scheduling.

Which dataset is used for the study?

The thesis uses the turbofan degradation dataset provided by NASA as the case study dataset.

How are machine learning models evaluated in the thesis?

Regression models and artificial neural networks are built and run on the dataset, followed by a comparative analysis of the results, supported by data visualization, feature selection, data treatment, and hyperparameter tuning.

Aug 5, 2026

Upload by Quinn