Remaining Useful Life Estimation Combining Two-Step Maximal Information Coefficient and Temporal Convolutional Network With Attention Mechanism

Yalan Jiang,Xianbo Wang,Chaoshun Li,Yujie Zhao,Zhixin Yang

doi:10.1109/access.2021.3052305

Abstract

Remaining useful life (RUL) estimation has received extensive attention in many fields, which provides improved decision-making for condition-based maintenance (CBM) and enhances the stability of engineered systems. However, the effective construction of the feature set and design of the high-accuracy prediction model are the main challenges in RUL estimation. Consequently, a novel RUL estimation method is proposed in this paper. Initially, to extract the deep characteristics of the raw sensor data, the modified ensemble empirical mode decomposition (MEEMD), combined with the correlation coefficient threshold to select the sensitive intrinsic mode function (IMF) components, is proposed to reconstruct more representative features. Additionally, maximal information coefficient-based two-step feature selection (TSMIC) method is used to select the optimal feature subset. Ultimately, temporal convolutional network with attention mechanism (TCNA) is proposed to capture long-term time-series information and achieves more precise prediction results. The proposed framework is verified through a case study on aircraft turbofan engines, and comparisons with other state-of-the-art methods are presented. The experimental results of the case study show the effectiveness and superiority of the proposed approach.

Highlights

With the rapid development of sensor technologies, telecommunications and computing systems, the industrial internet of things has seen wide application into prognostic and health management (PHM) of machinery in recent years
The proposed framework is implemented and evaluated on the degradation dataset of the aircraft turbofan engine, which was produced by the commercial modular aero-propulsion system simulation (C-MAPSS)
The experiment results show that the proposed data processing method modified ensemble empirical mode decomposition (MEEMD)-two-step feature selection method based on MIC (TSMIC) can effectively reconstruct the degradation features from raw sensor measurement data, and select the sensitive features which have high correlation with the target remaining useful life (RUL)

Summary

INTRODUCTION

With the rapid development of sensor technologies, telecommunications and computing systems, the industrial internet of things has seen wide application into prognostic and health management (PHM) of machinery in recent years. RNN, long-short term memory (LSTM) first proposed by [20], performs well in time-series prediction via reducing the difficulty of learning long-term dependencies by remembering historical information in the sensor data. To further improve the accuracy of network prediction and reduce the training time, the attention mechanism is introduced to filter information by calculating its contribution to improve the efficiency of the neural network [29], which has been widely applied and produced effective results in visual question answering [30], fine-grained visual recognition [31], VOLUME 9, 2021. Reference [34] combined BLSTM with attention mechanism to predict the voltage degradation of the proton exchange membrane fuel cells stack These researches suggested that improved results can be obtained by using attention-based neural network model compared to other ordinary deep learning models.

THE PROPOSED FRAMEWORK

MODIFIED ENSEMBLE EMPIRICAL MODE DECOMPOSITION

TWO-STEP MAXIMAL INFORMATION COEFFICIENT METHOD

THE TCNA PROGNOSTIC MODEL

COMPARISONS AND ANALYSIS

CONCLUSION