Multiobjective Collaborative Optimization of Systems of Systems

摘要

Concept studies for warship designs typically focus on ship performance characteristics by setting design goals for such things as speed, range, and cost. However, warships generally operate as part of a larger battle or strike group. Therefore, the designs should be evaluated as part of a system of multiple ship systems since designing each ship individually may result in underutilized and excess equipment and capability; in other words an inefficient design of the system of systems. This thesis examines the simultaneous design of several ships using the sea base concept as an example application of a network of ships working together. The number and characteristics of these ships determine the mission performance of the sea base. To properly design any of the sea base ships, the interrelationships must be included. A mission simulation is used to combine the performance characteristics of different ship designs into a single performance objective: the time to deliver a brigade size combat force to its assigned objectives. To enable the design of multiple ships, collaborative optimization, a multilevel optimization approach, was used to decompose the problem into individual ship design optimizations with system level interfaces controlled by a system of systems optimization algorithm. This allowed each ship to use techniques and algorithms best suited to reach an optimal design without impacting the design approaches used by the other ships. The classical collaborative optimization approach was relaxed to include multiple objectives such as performance and cost, thus developing a range of solutions which represent the tradeoff between these objectives.