A hybrid approach to space power control utilizing expert systems and numerical techniques

The feasibility of combining the advantages of conventional control processing techniques with an expert system for space power system control is discussed. The results reported here focus on diagnostic evaluation and control of specific components and assemblies in the Rocketdyne Space Power Laboratory. Operating data and failure mode analyses are used to formulate the initial object-oriented rule base for the expert system running under the Nexpert operating shell. Online tests in the laboratory environment have been encouraging, demonstrating that the interface necessary for cooperative operation between a central controlling processor and an expert system is feasible.<
>