Building parametric and probabilistic dynamic vehicle models using neural networks

During the past decade, the aircraft vehicle design process has undergone a major shift of focus from pure performance towards a balance between vehicle characteristics and cost, namely aordabilit y. In addition, accelerated advances in computing technology have helped render a complete parametric and probabilistic design process feasible. All of these changes have allowed more knowledge to be brought earlier into the design process, which helps designers make more informed and therefore better decisions, earlier in the design process. Computing power now allows extensive physics-based vehicle modeling early in the design cycle. A full non-linear six degree of freedom parametric dynamic vehicle model should be attainable as early as the conceptual design phase. Such a vehicle model would help understand the eects of design variables on vehicle characteristics and operation through analysis and simulation. Furthermore, probabilistic design methods allow for the proper treatment of uncertainty and delit y inherent in such a model. This paper formulates a framework to arrive at a conceptual non-linear six degree of freedom parametric and probabilistic dynamic vehicle model based on neural networks.