A multi-population firefly algorithm for dynamic optimization problems

In traditional optimization problems, problem domain, constraints and problem related data are assumed to remain stationary throughout the optimization process. However, numerous real life optimization problems are indeed dynamic in their nature due to unpredictable events such as due date changes, arrival of new jobs or cancellations. In the literature, a problem with one of these features is referred as dynamic optimization problem (DOP). In contrast to static optimization problems, in DOPs, the aim is not only to find the optimum of the current configuration of a problem environment, but to track and find the changing optima. The field of dynamic optimization is a hot research area and it has attracted a remarkable attention of researchers. A considerable number of recent studies on DOPs usually employs bio-inspired metaheuristic algorithms, which are efficient on a wide range of static optimization problems. In the present work, a multi-population firefly algorithm with chaotic maps is proposed to solve DOPs. The tests are conducted on the well known moving peaks benchmark problem. In regard to the results, the proposed algorithm is found as a promising approach for the present problem.