Hyperheuristics for Determination of Non-dominated Set of Public Service System Designs

A design of a public service system subjects to various objectives, which are usually in conflict. The most known pair of conflicting criteria is the system and fair criterion, where the system criterion expresses utility or disutility of an average system user and the criterion of fairness takes into account the access of the worst situated minority of the system users to service. A series of non-dominated system designs is important especially for the decision maker responsible form the final form of the system. In this contribution, we concentrated on study and construction of hyperheuristics assigned to the efficient determination of a non-dominated set of public service system designs, where the system and fair criteria are taken into account. The suggested hyperheuristic disposes with a list of subordinate heuristics with dynamically updated ranks depending on their previous success in improving quality of the non-dominated solution set. To explore properties of the employed subordinate heuristic, a series of numerical experiments with real-sized benchmarks has been performed and the obtained results are presented and discussed.