Abstract:

The paper deals with testing different strategies - Inertia Weight Strategies, Constriction Factors and Interpersonal Learning – of a Particle Swarm Optimization implemented in a simulation optimizer. The strategies were tested on five different discrete event simulation models reflecting real optimization problems in industrial companies. We specified different objective functions of discrete event simulation models considering the simulated system. We tested different settings of the strategies to reduce bad settings. We replicated optimization experiments with concrete optimization method parameters settings to reduce the randomness of the behaviour of the optimization method strategies. We evaluated the success of finding the optimum by the different PSO strategies according to other optimization methods - Random Search, Downhill Simplex, Hill Climbing, Tabu Search, Local Search, Simulated Annealing, Evolution Strategy, Differential Evolution and Self Organizing Migrating Algorithm.