Robust-MBDL: A Robust Multi-branch Deep Learning Based Model for Remaining Useful Life Prediction and Operational Condition Identification of Rotating Machines

The prediction of RUL has garnered significant attention from both academic researchers and industry professionals. This is because accurately predicting RUL can significantly enhance the effectiveness of predictive maintenance, leading to increased machine reliability and reduced incidences of failures and associated repair costs. Existing RUL prediction models generally fall within two primary categories: the model-based and data-driven approaches [8]. The model-based approach relies on a certain level of physical knowledge about machine degradation to predict RUL, such as employing theories of the Paris law for bearing defect growth [18] and reliability laws [42, 3, 44]. However, integrating such physical knowledge into models can be challenging, especially concerning complex machinery where such insights might not always be readily available.