Airframe Performance Optimization of Guided Projectiles Using Design of Experiments

Abstract

Performance optimization of guided, gun-launched projectiles is a difficult task due to nonlinear flight behavior, complex aerodynamic interactions, and unique engineering constraints. Historically, the design process for many smart weapons has been iterative in which a series of design improvements are made until performance requirements have been met. This paper presents an alternative formal methodology for smart weapons conceptual airframe design and optimization based on design of experiments. At the initial stage, a basic aerobody shape is defined along with candidate control actuators and associated design parameters. Based on a design of experiments, a kriging response surface is generated mapping design variables to performance criteria. Simultaneously, a neural network is trained to recognize unstable designs. Finally, a genetic algorithm determines the optimal projectile design with respect to a predefined cost function. By varying this cost function, a Pareto frontier of optimal designs can be generated reflecting performance tradeoffs. A detailed description of the methodology is given, along with an example in which the fin configuration of a projectile is optimized based on multivariate criteria that seeks to maximize range and impact velocity while minimizing angle of attack. Results show that the proposed automated optimization process is a feasible and valuable tool for smart weapons conceptual airframe design.