RUL Estimation Enhancement using Hybrid Deep Learning Methods

Abstract

The turbofan engine is one of the most critical aircraft components. Its failure may introduce unwanted downtime, expensive repair, and affect safety performance. Therefore, It is essential to accurately detect upcoming failures by predicting the future behavior health state of turbofan engines as well as its Remaining Useful Life. The use of deep learning techniques to estimate Remaining Useful Life has seen a growing interest over the last decade. However, hybrid deep learning methods have not been sufficiently explored yet by researchers.In this paper, we proposed two-hybrid methods combining Convolutional Auto-encoder (CAE), Bi-directional Gated Recurrent Unit (BDGRU), Bi-directional Long-Short Term Memory (BDLSTM), and Convolutional Neural Network (CNN) to enhance the RUL estimation. The results indicate that the hybrid methods exhibit the most reliable RUL prediction accuracy and significantly outperform the most robust predictions in the literature.

RUL Prediction Using a Fusion of Attention-Based Convolutional Variational AutoEncoder and Ensemble Learning Classifier

“…CNN architectures are designed to extract features through weight sharing filters and showed a noticeable improvement in prediction accuracy. Hybrid deep neural network models have also been reported in the literature [31], [32], [33] to leverage the power of different DL methods, which integrate CNN and LSTM models simultaneously to extract temporal and spatial features.…”

“…Deep learning models are full of hyperparameters in terms of architecture and training parameters (such as the number or type of layers and the learning rate). Their optimization by most of the reviewed papers are based on a trial-anderror approach [29], [15], [16], [17], [18], [22], [23], [30], [31], [32], [51], [33]. However, this approach can be timeconsuming and error-prone due to a lack of understanding of the impacts of parameters.…”

A method to estimate the remaining useful lifetime of a two-jaw parallel gripper based on experimental failure threshold data

“…Deep learning (DL) has attracted strong interest in Prognostics and Health Management (PHM) applications, due to its high processing power, automated feature learning capability, and problem-solving ability [28]. Several publications concur that models using DL approaches perform more accurately than those without any DL integration [29][30][31]. The potential, challenges, and future directions for DL in PHM have been published by Fink et al [32].…”