Preliminary design of unmanned aircraft using genetic algorithms and data mining

Aircraft design is a complex process involving multiple co-dependent design variables and many design decisions. For commercial aircraft design, this difficulty is offset somewhat by the wealth of knowledge available. Observing existing designs has provided useful empirical relationships and insights for the designer to apply, yielding a relatively well defined problem. The wide variety of configuration possibilities, mission profiles, and the relative lack of historical data leave the problem of unmanned aerial vehicle (UAV) design less defined. The purpose of this research was to develop a robust optimization package for UAV design using data mining to aid configuration decisions and to develop empirical relationships applicable to a wide variety of mission profiles. An optimization software package was developed using a Genetic Algorithm (GA) and Data Mining. The algorithm proved successful in carrying out the preliminary design phase of a number to test cases similar to existing UAVs. Designs produced by the algorithm promise improved performance flight performance relative to existing systems, and reduced development time when compared with conventional design methodology. Future work will introduce high fidelity analysis to the framework developed in this research.