Incorporating Transfer Learning in Engine Health Assessment for Accurate Remaining Useful Life Prediction

Aircraft engines are critical, therefore any move to make them safer, more reliable, and fuel-efficient is highly encouraged. To solve the challenges of flight safety and maintenance expense during aircraft engine operation, a prognostics and health management system is deployed. This system is primarily concerned with flaw detection, health evaluation, and life prediction. Making decisions about aircraft engine operation and maintenance is strongly reliant on estimating the remaining useful life (RUL). This study provides an ensemble transfer learning model for RUL prediction using the C-MAPSS turbofan engine dataset. Pre-processing procedures such as filtering and normalization are used to improve the dataset. The proposed ensemble model for RUL prediction is then trained on the processed data. Some of the well-known transfer learning models that are compared include Xception, ResNet-50, and VGG-16. The ensemble model outperforms the other models, with the highest R2 value of 0.9901. Furthermore, the ensemble model is very effective; it can predict RUL in just 37 seconds, which suggests that it could be useful for keeping an eye on engine health and making decisions about aircraft.