Comparison of a physical model and a machine learning approach for more accurate assessment of fuel efficiency measures

摘要

Using full flight data of a commercial transport aircraft the performances of different approaches to assess the influence of fuel efficiency measures on the actual fuel consumption are tested and compared to a standard method. One algorithm is using a pure statistical machine learning approach. The other is making use of a algorithm, which is combining a physical model of the aircraft with a parameter estimation method to determine aircraft specific coefficients based on normal flight data.