A Bidirectional Long Short-Term Memory Autoencoder Transformer for Remaining Useful Life Estimation

Abstract

Estimating the remaining useful life (RUL) of aircraft engines holds a pivotal role in enhancing safety, optimizing operations, and promoting sustainability, thus being a crucial component of modern aviation management. Precise RUL predictions offer valuable insights into an engine’s condition, enabling informed decisions regarding maintenance and crew scheduling. In this context, we propose a novel RUL prediction approach in this paper, harnessing the power of bi-directional LSTM and Transformer architectures, known for their success in sequence modeling, such as natural languages. We adopt the encoder part of the full Transformer as the backbone of our framework, integrating it with a self-supervised denoising autoencoder that utilizes bidirectional LSTM for improved feature extraction. Within our framework, a sequence of multivariate time-series sensor measurements serves as the input, initially processed by the bidirectional LSTM autoencoder to extract essential features. Subsequently, these feature values are fed into our Transformer encoder backbone for RUL prediction. Notably, our approach simultaneously trains the autoencoder and Transformer encoder, different from the naive sequential training method. Through a series of numerical experiments carried out on the C-MAPSS datasets, we demonstrate that the efficacy of our proposed models either surpasses or stands on par with that of other existing methods.

A Two-Stage Attention-Based Hierarchical Transformer for Turbofan Engine Remaining Useful Life Prediction

“…Table 5 shows the comparison results. [56], GCT [36], DCNN [57], ELSTMNN [58], DATCN [59], AGCNN [60], BiLSTM attention model [61], DAST [46], DLformer [37], 1D-CNN-LSTM [62], CNN-LSTM-SAM [63], and BiLSTM-DAE-Transformer [42]. As presented in Table 5, the proposed STAR framework consistently outperforms existing RUL prediction models across all the datasets, showcasing its superior predictive capabilities.…”

“…Rooted in the Transformer architecture and incorporating a gated mechanism, the BGT model effectively quantifies both epistemic and aleatory uncertainties and providing risk-aware RUL predictions. Most recently, Fan et al [42] introduced the BiLSTM-DAE-Transformer framework for RUL prediction, utilizing the Transformer's encoder as the framework's backbone and integrating it with a self-supervised denoising autoencoder that employs BiLSTM for enhanced feature extraction.…”

“…To benchmark its effectiveness, we compare the proposed model against a suite of existing methods widely recognized in the field. These methods include BiLSTM [57], the gated convolutional Transformer (GCT) [36], DCNN [58], ELSTMNN [59], DATCN [60], AGCNN [61], BiLSTM attention model [62], DAST [46], DLformer [37], 1D-CNN-LSTM [63], CNN-LSTM-SAM [64], and BiLSTM-DAE-Transformer [42]. Table 5 shows the comparison results.…”

Robust-MBDL: A Robust Multi-Branch Deep-Learning-Based Model for Remaining Useful Life Prediction of Rotating Machines

“…The experimental study confirmed that the proposed model outperforms the CNN, LSTM, and Auto-encoder models. In [18], the raw data are first denoised using a Bi-LSTM autoencoder before being passed into the Transformer encoder for RUL prediction. The denoising step's effectiveness was confirmed through various experiments.…”

“…In the literature, several DL models have been proposed to estimate the RUL of rotating machines. These models often consist of simple neural network architectures, including a series of several Long Short-Term Memory (LSTM) or convolutional layers [3,17,18]. The convolutional layers are directed towards the identification of spatial dependencies within time series data, while the LSTM layers excel at identifying and capturing long-term correlations.…”