Research on Remaining Useful Life Prediction Method of Rotating Machinery Based on the Fusion of GRU and Transformer

The remaining useful life (RUL) of large rotating machinery has an important significance for prognostic health management (PHM) tasks and safety performance evaluation. In order to predict the remaining useful life of rotating machinery, this paper combines gated recurrent network (GRU) and Transformer network to process time series data. It introduces a scale aware function enhancement (SFE) module to extract features from the dataset, achieving accurate prediction of the remaining useful life. Specifically, the Transformer is the main component of the model, and its role is to capture short-term and long-term dependencies in time series. Since the features output is processed by the self-attention mechanism in the Transformer model, and it is not sensitive to the local feature context of the data, the GRU model is adopted, enabling the model to have the ability to include data context at each time step. Meanwhile, the scale aware function enhancement (SFE) module is employed to effectively extract features. Experiments analysis using the CMAPSS dataset are conducted and the experimental results demonstrate that the predictive performance of the model is better than other existing methods.