Bayesian Networks for US Army Electronics Equipment Diagnostic Applications: CECOM Equipent Diagnostic Analysis Tool, Virtual Logistics Assistance Representative
Soldiers in war zones depend on complex weapon systems and advanced technologies. Moreover, in combat conditions, the resources available to support operation and maintenance of these systems are minimal. An important constraint is time, as transporting technical experts to and from combat outposts is cumbersome. Also, technical experts face the same battlefield risks as soldiers. CECOM has developed a suite of systems, named Virtual Logistics Assistance Representative (VLAR), to maximize self-reliance at combat outposts. In this paper, we discuss the operations research methods that underpin VLAR, at the heart of which lies Causal Bayesian belief networks. We outline the methods we use to translate scientific theory into accessible applications, the processes that optimize Army tactical electrical power grids, and the intuitive soldier interface that is common to all VLAR products. VLAR is changing the Army’s sustainment paradigm by applying artificial intelligence to equipment diagnostics and electrical power grid optimization. Through the end of 2015, we estimate that VLAR and Headquarters Fuel Optimization have saved the Army $35.4M from an investment of $9.8M, a benefit to cost ratio of 3.6. During this time the elimination of approximately 360,000 helicopter or ground vehicle movements has greatly reduced the risk of improvised explosive device and attacks, preventing roughly 4,500 casualties (killed and wounded). We project additional savings of $222M from an investment of $60M by the end of 2020, a benefit to cost ratio of 3.7. During this period, we project the elimination of 900,000 helicopter or ground vehicle movements and 11,000 transport casualties an indirect savings of $22B in long-term medical costs. (Bilmes, 2013).