Leveraging Ground Test Data for Aero-Engine Thrust Estimation

Accurate aero-engine thrust estimation plays an important role in direct thrust control. This work proposes to leverage the ground test data to estimate the thrust with an artificial neural network. The Savitzky Golay method is used to denoise the data collected by sensors. The random forest and Pearson correlation coefficient methods are combined to reduce the dimension of the data. Considering the limitation of sensors on flight comparing to the ground test, we propose to use the on flight sensor data to estimate the missing data first, and then integrate them as inputs to estimate the thrust. The results of case study validate the effectiveness of our proposed model.