Optimal Design of Uncertain Complex Dynamical Systems.

摘要

Recent advances in simulation and computation call for innovative modeling and design approaches for engineered systems that can account for the complexities and uncertainties in the system description in design processes. In this research we consider the problem of optimal design of uncertain complex systems. A framework has been developed for the optimal design of complex dynamical systems that incorporates model uncertainties directly into the design objective. A class of systems that exhibit complex behavior that can be approximated by a hybrid system is considered. The discrete behavior evolves on a slow time scale and can be modeled as a hidden Markov process. A model reduction technique that captures all critical aspects ol the discrete component has been developed. Data driven identification techniques have been developed for identification of the hidden Markov model. Furthermore, system identification is used for the characterization of fast time scale dynamics of the hybrid components resulting in reduced models that are suitable for design. The novelty of the approach lies in hybrid system modeling approach.