Attention-Based Bidirectional LSTM-CNN Model for Remaining Useful Life Estimation

In many industries, prognostic health management (PHM) technology has become important as a key technology to increase reliability and operational efficiency. Recently, several methods using a deep learning architecture to estimate the remaining useful life (RUL) as a part of the PHM have been presented. However, the limitation of existing methods is that they do not explicitly capture the relationship among different time sequences, which reduces the accuracy of RUL estimation. This paper proposes a novel RUL estimation algorithm using the attention mechanism to solve this problem. The proposed method applies scaled dot product attention to the encoder and the decoder consisting of long short-term memory, convolutional neural network and fully connected layer. The encoder applies self-attention to extract the association between time sequences, and the decoder extracts the association between the target RUL value and the time sequences using the representative vector of the RUL. Therefore, the proposed model has better performance to capture the long-term dependency in the sequence data and outperforms other state-of-the-art models in the experimental results. In addition, the extracted attention map shows that our model has better interpretability for RUL estimation.

I. Introduction

In recent years, prognostic health management (PHM) has become important to improve the operational efficiency, reliability, and system performance. PHM continuously monitors the system operation, and diagnoses abnormal signs, when failure levels or unusable conditions occur. With the PHM technology, condition-based predictive maintenance can be performed only when necessary, and maintenance costs can be greatly reduced. In PHM, prognostics estimate the remaining useful life (RUL) [1] for the system to perform its intended function. The importance of RUL estimation in many fields, such as the aircraft industry, medical equipment and power plants, has encouraged researchers to develop a variety of RUL estimation approaches [2]-[6]. Recently, deep learning has shown remarkable achievements in image recognition and speech recognition [7], [8]. Deep learning is characterized by a deep architecture in which several layers are stacked to capture representative information from raw input data [9]. This characteristic of deep learning has great potential in matching original data and RUL. Therefore, it is very useful to be used in the RUL estimation.