Dynamic Differential Evolution with Difference Mean Based Perturbation

Optimization in a dynamic environment is a real challenge owing to the multimodality, high complexity, and ruggedness of the functions involved. Tracking the global optima in such a dynamically changing landscape is called Dynamic Optimization Problem(DOP). This paper aims at modifying the popular DOP handling technique Dynamic Differential Evolution( DynDE) by introducing a unique scheme named Difference Mean Based Perturbation (DMP) technique apart from classical DE that greatly enhances the diversity of the algorithm and offers a scope of thorough search of area in the vicinity of the current best. The proposed algorithm is hence addressed as DynDE-DMP. The other features of DynDE-DMP include an aging mechanism, an exclusion principle and a cluster based retention strategy. Performance of DynDE-DMP has been tested over the suite of benchmark problems used in Competition on Evolutionary Computation in Dynamic and Uncertain Environments, held under the 2009 IEEE Congress on Evolutionary Computation (CEC) and compared with six state-of-the-art EAs. The comparison results reflect the effectiveness of the DMP scheme thus establishing the proposed approach, a successful optimizer in Dynamic Environments.