An Environmental Adaption Method with real parameter encoding for dynamic environment

Abstract

A dynamic version of Environmental Adaption Method (EAM) is proposed in this paper. Environmental Adaption Method for Dynamic Environment (EAMD) is an improvement over EAM, which works in dynamic environment with real valued parameters. Unlike EAM the theory of this algorithm is based on adaption of species in dynamic environment which gradually becomes more verse and deadly for their denizens. The species which are able to adapt in the changing environment, improves their fitness value by enhancing their phenotypic structure in the upcoming generations. Sudden and gradual dynamic changes in the environment assist species to converge towards the optimal fitness. Unlike EAM, EAMD is suitable for both unimodal and multimodal problems without the need of an alteration operator as there is enough diversity since the adaption is randomized, i.e. each possible solution can adapt anywhere within the search space. EAMD is compared with various algorithms tested on 24 benchmark functions against the Black Box Optimization Benchmarking (BBOB) test-bed at different dimensions with very promising results and EAMD shows its superiority over other state-of-the-art algorithms.