Metric-Based Heuristic Space Diversity Management in a Meta-Hyper-Heuristic Framework

This paper investigates various strategies for the management of heuristic space diversity within the context of a meta-hyper-heuristic algorithm. In contrast to all previously developed heuristic space diversity management strategies, this paper makes use of a heuristic space diversity metric to monitor heuristic space diversity throughout the optimization run and trigger the need for increased or decreased heuristic space diversity. Three different heuristic space diversity management strategies are evaluated. Maintaining a high level of heuristic space diversity throughout the optimization run is shown to be the best performing strategy. Good performance is also demonstrated with respect to a state-of-the-art multi-method algorithm, another successful diversity controlling meta-hyper-heuristic and the best-performing constituent algorithm.