RIASSUNTO
CAPSO is a parallelized hybrid optimization system designed for solving multiobjective problems. CAPSO combines elements from Cultural Algorithms (CAs), Particle Swarm Optimization (PSO), and Vector‐Evaluated Genetic Algorithms (VEGAs). CAPSO works by dividing a large search space between multiple particle swarms joined by the sharing of CA knowledge among themselves. In this paper, we investigate the relative contribution of different CA knowledge sources in the deployment of PSO swarms in a search for the Pareto Optimum in Constrained multiobjective optimization problems. We show that depending on certain symmetries in the search space, certain categories of knowledge sources are able to dominate others in the search process. While exploratory knowledge sources tend to dominate search in unconstrained problems, exploitative knowledge sources are able to exploit search space patterns and symmetries in constrained problems. The dominance hierarchy that emerged for each of the example problems was different for each. That suggests the flexibility of such a knowledge‐driven approach to handle a variety of constrained problem types.