Capability Test Design and Analysis at the Naval Postgraduate School

摘要

The Joint Test and Evaluation Methodology project developed in collaboration with the Naval Postgraduate School's Simulation Experiments and Efficient Designs Center for Data Farming and Training and Doctrine Command Analysis Center-Monterey to enhance the design of experiment modeling and analysis approaches for testing in a joint environment. This article discusses the applied research conducted in this area over the past 3 years as well as its application to joint test and evaluation methodology test events. Discoveries involving enhanced data farming techniques and technology applications have proven to be catalysts for the test and evaluation of complex adaptive systems. Hybrid design of experimental models for large factor test designs, e.g., fractional factorial controlled sequential bifurcation, resolution five fractional factorial, nearly orthogonal Latin hypercube, have demonstrated success in refining robust joint test spaces. Innovative application of analytical models and methodologies, e.g., advanced response surface methodology and classification and regression tree, have improved our ability to analyze critical operational issues with joint impact involving multiple responses. Agent-based model simulation prototypes, e.g., Tester, Map Aware Non-uniform Automata, Pythagoras, have been modified and/or developed by our academic and government partners to enable enhanced test design and evaluation of capabilities in a joint environment. Proof of concept efforts in this collaboration have included international data farming workshop events, where various techniques and tools have been explored for use in testing in a joint environment. Key research techniques and selected results are presented in the context of a use case that is based upon joint test and evaluation methodology test events.