Remaining Useful Life Prediction of Aeroengine Based on Ghost Approach

Aeroengine is an important part related to aircraft flight safety. Accurate prediction of the remaining useful life for the engine is of great significance in real life. Aiming at the current problem of limited real-time RUL prediction, a novel prediction model based on the Ghost method is proposed. The Ghost method can take advantage of the redundancy of feature maps in CNN, and use simple operations to obtain ghost feature maps, instead of redundant feature maps in traditional CNN. Therefore, it can fully reduce the number of internal trainable parameters of the model, and accelerate training process. Validated using the C-MAPSS data set, the results show that, compared with the existing methods, the Ghost model has the smallest root mean square error and scoring function. While maintaining high-precision RUL prediction performance, the training speed of the model is faster than other methods.