Multiple fault diagnosis using factored evolutionary algorithms.

When supporting commercial or defense systems such as aircraft avionics, guidance and control, electronic warfare, or propulsion systems, a perennial challenge is providing effective test and diagnosis strategies to minimize downtime, thereby maximizing system availability. One can argue that one of the most effective ways to maximize downtime is to be able to detect and isolate as many faults that either exist or are emerging in a system at one time as possible. This is referred to as the "multiplefault diagnosis" problem, and it is a problem that is known to be computationally intractable (i.e., NP-complete) [1]. While several tools have been developed over the years to assist in addressing the multiple-fault diagnosis problem, considerable work remains to provide the best diagnosis possible given the information collected through observations, gripes, test results, and historical data. Recently, a new model for evolutionary computation has been developed called the "Factored Evolutionary Algorithm" (FEA) [2]. In FEA, a target optimization problem is broken down into subproblems that exhibit some kind of overlap. Then the optimization algorithm of choice (e.g., simulated annealing, genetic algorithm, particle swarm optimization) is applied to each of the subproblems, and the subproblems periodically share information with neighboring subproblems along the points of overlap. [ABSTRACT FROM AUTHOR]