PROJECT TITLE :

Machine Learning and Mass Estimation Methods for Ground-Based Aircraft Climb Prediction

ABSTRACT:

During this paper, we apply machine learning ways to enhance the aircraft climb prediction in the context of ground-based applications. Mass may be a key parameter for climb prediction. As it's considered a competitive parameter by several airlines, it is currently not available to ground-based trajectory predictors. Consequently, most predictors today use a reference mass which will be totally different from the actual aircraft mass. In previous papers, we have introduced a least squares method to estimate the mass from past trajectory points, using the physical model of the aircraft. Another mass estimation technique, primarily based on an adaptive mechanism, has also been proposed by Schultz et al. We tend to now introduce a brand new approach, in which the mass is considered the response variable of a prediction model that is learned from a set of example trajectories. This machine learning approach is compared with the results obtained when using the base of aircraft data (BADA) reference mass or the 2 state-of-the-art mass estimation ways. In these experiments, nine totally different aircraft types are thought of. When compared with the baseline methodology (respectively, the mass estimation ways), the Machine Learning approach reduces the RMSE (Root Mean Square Error) on the expected altitude by at least fifty eight% (resp. 27%) when assuming the speed profile to be known, and by a minimum of 29% (resp. seventeen%) when using the BADA speed profile apart from the aircraft varieties E145 and F100. For these types, the observed speed profile is much from the BADA speed profile.