Aircraft Mission Analysis Enhancement by Using Data Science and Machine Learning Techniques

As air-traffic demand continues to grow, it is expected that there will be a growing need for optimizing fuel consumption to airlines. To that end, it is prerequisite to estimate fuel consumption as accurate as possible. However, most of the aircraft operation datasets have been elusive due to proprietary purposes. Under these circumstances, a comprehensive software dubbed the Aviation Environmental Design Tool (AEDT) has been prevalently used by many aerospace engineers to calculate the fuel consumption. It is highly hypothesized that, besides, if the AEDT could collaborate with reliable weather information, its modeling fidelity would be enhanced. In this paper, we proposed a novel aircraft mission analysis framework by incorporating data-driven approaches with the AEDT and state-of-the-art machine learning (ML) techniques to improve the accuracy of aircraft mission analysis. As the source of weather information, the world-wide weather dataset called MERRA-2 was regressed by a Support Vector Machine (SVM) along the time and three-dimensional coordinates in the entire US territory. The created SVM model successfully provided continuous behavior of weather, showing a good agreement to the reference data. As the final research, a four dimensional (4-D) flight trajectory of operations in several sampled airports has been retrieved from external public databases and integrated with corresponding weather information extracted from the SVM model. Finally, it was observed that the collaboration of the SVM weather model and the AEDT precisely matched the reference data. The accomplishments of this research recommend that researchers conduct further study on the highly capricious behavior of weather with the power of the data-science and machine learning technology