Development of an Adaptive Aero-Propulsive Performance Model in Cruise Flight – Application to the Cessna Citation X

To accurately predict the amount of fuel needed by an aircraft for a given flight, a performance model must account for engine and airframe degradation. This paper presents a methodology to identify an aero-propulsive model to predict the fuel flow of an aircraft in cruise, while considering initial modeling uncertainties and performance variation over time due to degradation. Starting from performance data obtained from a Research Aircraft Flight Simulator, an initial aero-propulsive model was identified using different estimation methods. The estimation methods studied in this paper were polynomial interpolation, thin-plate splines, and neural networks. The aero-propulsive model was then structured using two lookup tables: one lookup table reflecting the aerodynamic performance, and another table reflecting the propulsive performance. Subsequently, an adaptative technique was developed to locally and then globally, adapt the lookup tables defining the aero-propulsive model using flight data. The methodology was applied to the Cessna Citation X business jet aircraft, for which a highly qualified level D research aircraft flight simulator was available. The results demonstrated that by using the proposed aero-propulsive performance model, it was possible to predict the aerodynamic performance with an average relative error of 0.99%, and the propulsive performance with an average relative error of 3.38%. These results were obtained using the neural network estimation method.