| Following the advances in electronics and communications technology in the last three decades, a new paradigm for large-scale dynamic systems emerged. In this paradigm, groups of independent dynamic systems, such as unmanned air vehicles or spacecraft, act as a cooperative unit for a diverse set of applications in remote sensing, exploration, and imaging. These systems have been envisioned to provide highly flexible and reconfigurable structures that use individual autonomy to respond to changing environments and operations.; The main aim of this research has been to design methods and algorithms to enable efficient operations for such large-scale dynamic systems when a centralized decision-maker cannot or does not exist. Towards this end, a decentralized optimization method and a coordination algorithm have been developed.; The decentralized optimization framework exploits a structure inherent in the problem formulation in which each decision maker has a mathematical model that captures the local dynamics and interconnecting constraints. A globally convergent algorithm based on sequential local optimizations is presented. Under the assumptions of differentiability and the linear independence constraint qualification, we show that the method results in global convergence to feasible Nash solutions that satisfy the Kuhn-Tucker necessary conditions for Pareto-optimality. Analysis of the second order sufficiency conditions provide insight to structures and solutions with strong local convexity or weak interconnections which guarantee local Pareto-optimality.; This methodology is applied to decentralized coordination problems from the aerospace and the operations research fields. We demonstrate the algorithm numerically via a multiple unmanned air vehicle system, with kinematic aircraft models, coordinating in a common airspace with separation requirements between the aircraft. In addition, analytic solutions are provided for decentralized inventory control in simple supply-chain networks. |
