Multi-objective Optimization of Network of Systems

Quek Kwang Ti

Abstract

There are many systems around us that can be broken down to sub-systems connected in a grid square structure. The sub-systems may be connected using physical means (relay electrical circuits, vehicle mechanism) or non-physical means (telecommunication network for relay of information).

A method for solving the multi-objective optimization problem for a grid network of solutions via redundancy allocation is proposed. The key feature of this algorithm is the conversion of objectives other than maximizing of reliability to resource limited constraints and the single objective problem is then solved iteratively for various combinations of the limits of the resources using the Dinghua Algorithm. An optimal solution is found for each of these iterations. The heuristic algorithm is able to achieve results close to that of the optimal solution using much lesser time than using exact method via full enumeration of the possible solutions. From the set of optimal solutions obtained, a Pareto optimal set of solutions can be obtained. The user may have subjective views about the importance of the various objectives. Based on the set of Pareto optimal solutions, using decision model like AHP, the user can make decisions on the "best" redundancy levels of the sub-systems in the network.