Improving Aircraft Engine Remaining Useful Life Prediction Scheme Based on Handcrafted Features and Neural Network Integration

The prediction of Remaining Useful Life (RUL) serves as a key indicator in the prognostics and health management of industrial equipment, holding significant implications for enhancing the availability of industrial equipment and reducing lifecycle costs. The common scheme based on the neural network especially for the Long Short-Term Memory (LSTM) networks can achieve satisfactory results in RUL prediction, while they are incapable of actively selecting key features and efficiently extracting degradation information embedded in the data. In this paper, we propose a new hybrid model based on LSTM networks. Here, the Convolutional Neural Networks (CNN) is employed to extract key features, then a Bidirectional Long Short-Term Memory (BiLSTM) is utilized to capture long and short-term dependencies within the features. Besides, the attention mechanism is introduced to be applied to assign different weights to select important parts. In this way, by integrating the achieved statistical features, the accuracy of aircraft engine RUL prediction can be effectively enhanced. And, by experiment, our method is verified by carrying out on the C-MAPSS dataset. The corresponding results reveal that our approach is superior to the other reported neural network-based scheme, and it can be considered as one of the most promising schemes using for the RUL prediction of aircraft engine.