Diagnostic information fusion: requirements flowdown and interface issues

In this paper we examine the implementation of a real-time, diagnostic, information fusion algorithm, which achieves a more accurate diagnostic estimate of system health than individual estimates produced by heterogeneous collection of diagnostic tools. We discuss an approach that both decomposes performance requirements (through a flowdown) into the design and accommodates the differences among the interfaces of a heterogeneous collection of diagnostic tools. We focus in particular on issues which arise from an implementation which monitors system health in real-time and do not consider an off-line post-processing implementation. Through a systematic flowdown of performance requirements into the system design, the combined effects of all the diagnostic tools are considered in order to achieve the desired high level of accuracy for each fault of interest