Polynomial Metamodelling of Computer Simulation for UAV Design

Statistical technics and regressions methods, has gained more attention and developments in the field of machine learning, giving rise to new approaches and tools. In the case of aeronautical engineering, and in similar industries, these innovative technics aid to efficiently analyze large amount of data and achieve feasible designs. Here a Polynomial model construction is investigated. It is shown that combining cross validation and statistical tests methods results in satisfactory model accuracy with less tremendous computations. Some post processing is then used to locate feasible design variables ranges for longitudinal static stability of an unmanned aerial vehicle (UAV).