A Hybrid Prognostic Approach Based on Deep Learning for the Degradation Prediction of Machinery

Remaining useful life (RUL) prediction is of great significance for prognostic and health management (PHM) as it can achieve more reliable effective maintenance strategies. With the advances in field deep learning, data-driven methods have provided promising results. Hence, this research presents a approach based on learning models predicting RUL mechanical systems effectively. Multiple separable convolution layers, bidirectional Long Short-Term Memory (LSTM) layer, fully-connected layers (FCL) are included proposed network, named SC-BLSTM, to accomplish accurate from raw degradation data acquired by different sensors. The SC-BLSTM aims learn complex nonlinear features input capture temporal dependencies learned features. presented tested verified turbofan engines (C-MAPSS dataset) NASA. result demonstrated that able compared with some existing models.