A full-featured cooperative coevolutionary memory-based artificial immune system for dynamic optimization

In this paper, a novel cooperative coevolutionary memory-based artificial immune system enhanced by a new clonal selection algorithm is proposed for dynamic optimization problems. In the proposed algorithm, the whole n-dimensional population is decomposed into n one-dimensional subpopulations. Then, each subpopulation is evaluated separately using a set of context vectors called short-term memory. Also, inspired by the production of new cells in bone marrow, each subpopulation is divided into multiple regions to track and locate multiple optima cooperatively. This division helps the algorithm exploit search space effectively. Additionally, inspired by the immune memory concept, a memory-based approach called long-term memory is proposed to store and retrieve essential information when a fitness change occurs. Furthermore, a new clonal selection method, a combination of negative selection and clonal selection mechanisms, is proposed. This proposed algorithm is faster than the basic clonal selection algorithm. Finally, compared to other immune-based algorithms, which usually are implemented based on one or two qualities of the biologic immune system, the proposed approach exploit almost all immune qualities. Several experiments are conducted on different configurations of the moving peaks benchmark to examine the efficiency of the proposed method. The experimental results confirm that the proposed method is competitive with other state-of-the-art algorithms to optimize dynamic problems. •Combining of negative selection and clonal selection mechanisms, for increasing the convergence speed.•Producing the uniform random B-cell behavior simulation in bone marrow, for increasing diversity.•Simulating Memory-cell, to avoid forgetting the optima.•Scaping from local minima by multi epitope simulation and search space division.