Building Requirements-Flow Models Using Bayesian Networks and Designed Simulation Experiments
Terrill Hurst
Raytheon Missile Systems
Jarom Ballantyne
Raytheon Missile Systems
Allan Mense
Raytheon Missile Systems
This paper describes a method for constructing and evaluating Bayesian networks (BNs) as a logically consistent model of a set of derived system requirements. A BN consists of (1) a directed acyclic graph, (2) a set of fully defined states for each node in the graph, and (3) a conditional probability table (CPT) for each of the nodes. Designed simulation experiments are central to the construction and evaluation of BN models to predict system performance given a set of subsystem requirements. Probability estimates in each CPT are made by mining data from simulation experiments on prior systems and using engineering judgment to alter the CPT entries and explore options for satisfying top-level requirements. Sensitivity analysis is conducted to prioritize and assign values to each requirement. Benefits of using Bayesian networks are reported, including the ability to (1) analyze design margin, (2) allocate subsystem tolerances, (3) estimate the achievable upper bound on system performance for a proposed design improvement, and (4) integrate results from designed simulation experiments involving multiple subsystems.