Bi-level optimisation using genetic algorithm

In most real-life problems such as rolling system, design decision-making can be hierarchical and the search space is unknown. Bi-level optimisation problem (BLP) is an operation research technique for solving real life hierarchical decision-making problems. There are a number of different algorithms developed based on classical optimisation methods to solve different classes of the BLP problems where the search space is known. There also exist a number of problems in the BLP which current algorithms are not sufficiently robust to solve. In this paper, a bi-level genetic algorithm (BiGA) is proposed to solve different classes of the BLP problems within a single framework. BiGA is an elitist optimisation algorithm developed to encourage limited asymmetric cooperation between the two players. The performance of the algorithm is illustrated using test functions. The results suggest that BiGA algorithm is robust to solve different classes of the BLP problems and demonstrates potential for real life problems.