RUL Prediction of Turbofan Engine Based on WGAN-Trans Under Small Samples

The prediction of the remaining useful life of the turbofan engine of an aircraft is a very important part of its PHM. With the arrival of the era of Big data, data driven RUL prediction methods are gradually emerging, however, in the actual industry, it is often difficult to collect sufficient data to predict the RUL under deep learning which means there will be a problem of small samples. Moreover, in the early stage of degradation, it is difficult to accurately classify the health status of equipment, resulting in the inability to provide early warning before the rapid degradation of equipment performance, thus losing the significance of developing maintenance strategies in advance. To solve this problem, this paper proposes a WGAN-Trans model, which expands the data set by improving the Gener-ative adversarial network, uses the one-dimensional Convolutionl Neural Network to divide the health state, and finally uses the Transformer model to predict the RUL, which is verified on the commercial aircraft turbofan engine data set provided by NASA. The results show that the proposed model has good performance.