Integrating Support for Statistical Design Methods within Missile 6DOF Simulation Experiments

Abstract

This paper describes a strategy for integrating statistical experimental design methods (SDM) within a generic 6-degree-of-freedom (6DOF) missile simulation called genSim. Besides the basic framework and software tools, this integration requires training and ongoing collaboration with users so that they increase proficiency in conducting simulation experiments using SDM. The language and principles of SDM are team-taught with practitioners, who present actual case histories to illustrate application of SDM to 6DOF simulation. An open, modular course structure encourages students to develop skills and judgment for applying what they learn to their current needs, to improve course content, and collaboratively to enhance the infrastructure for future simulation experiments. 1 Implications of this trend include the requirement to conduct early high-fidelity 6DOF simulations or reduced-fidelity simulations that are based on quantifiably justified assumptions. The U.S. Armys Common Simulation Framework (CSF) was created to save time developing simulations by incorporating re-useable components. 2 At Raytheon Missile Systems, a CSF- based generic simulation framework called genSim is being used as the basis for creating program-specific 6DOF simulations. 3,4 Usage and enhancements of genSim are increasing, fostered by a CSF users group and a core genSim support team. The presence of an open framework invites collaborative expansion of initial capabilities. One such area of expansion is the support of statistical design methods (SDM) for simulation experiments. Initial genSim support for SDM included sweeps of input parameters (factors) and Monte Carlo analysis. Given a set of discrete values or probability distributions for factors, these functions are used to predict values and/or probability distributions for missile performance metrics (responses), such as miss distance. As genSim usersconfidence in SDM increases, the methods can be used in the process of simulation verification, validation, and accreditation (VV&A). An expanding list of desirable SDM features evolves as userscompetence increases. This paper describes elements of the strategy developed at Raytheon Missile Systems to integrate SDM within the 6DOF simulation process, including performance assessment and VV&A. The strategy acknowledges the pervasive use of Monte Carlo analysis and seeks to establish SDM as a complementary capability. Familiarity with the language of statistics, facility in making and checking assumptions, and computing and interpreting statistical properties, are all critical to bringing SDM into the 6DOF simulation context. It became clear that a training course could best meet these needs, and thus a project was authorized to develop and deliver this course. In addition to educating genSim users, the course serves as a forum for discussing how to improve genSim and to enhance the infrastructure to integrate SDM within simulation experiments. Section II briefly reviews the genSim framework and capabilities that were initially available for SDM. Section III describes several issues that arise when applying traditional SDMoriginally developed for real-world experimentswithin the 6DOF simulation context. Section IV describes the approach that was used to create a training course, called DASE, Design & Analysis of Simulation Experiments. Section V provides an example of the content of one of the DASE modules. Finally, Section VI provides conclusions on progress to-date and ideas for how the integration strategy will be applied in future simulation experiments. 1