Application of Robust Design Techniques to the Parallel Design of Engineering Systems

Abstract

The problem of designing complex systems, with performance specifications from multiple disciplines, was cast within the framework of robust control design in a previous publication. In that work, the overall system design problem was decomposed into several subsystem design problems. Robust controller analysis techniques were used to determine limits on the magnitudes of the uncertain parameters that would enable the design of the subsystems to proceed in parallel. In this paper, we extend that previous work, formulate the subsystem design problems in a robust controller design framework and use nonlinear optimization to design subsystems that are robust with respect to the uncertainties arising from designing the subsystems in parallel. This requires that the uncertain parameters for a given subsystem include, in addition to its own uncertain design parameters, the design variables and outputs from other subsystems that are needed in that subsystem analysis. Incorporating the uncertainties arising from parallel subsystem design into robust controller design framework will ensure that, upon assembly of the subsystems, the system-level performance requirements are met. The results are illustrated using a simple example representing a corner-car model.