Simulation-Based Ant Colony Optimization for Complex System Configuration Problems

Complex system configuration problems are the problems of appropriately assigning system parameter values for optimizing some aspect of complex system performance. In this paper, we first cast complex system configuration problems as mixed-variable parameter optimization problems where mensurable system simulation responses are used for evaluation. Then we present a simulation-based ant colony optimization algorithm (sACO(MV)) to tackle the problems. In sACO(MV), the decision variables of the complex system configuration problems can be clearly declared as continuous, ordinal, or categorical and let the algorithm treat them adequately. Finally, sACO(MV) is tested on mixed-variable complex engineering system configuration problems. The effectiveness and robustness of sACO(MV) are demonstrated by the comparisons with results from the literature.