Cloud droplets evolutionary algorithm on reciprocity mechanism for function optimization

Abstract

For the problems of solving difficult problems in evolutionary algorithms such as easily falling into local optimum, premature convergence because of selective pressure, a complex and larger calculation and a lower accuracy of the solution, this paper proposes cloud droplets evolutionary model on reciprocity mechanism (CDER). The main idea of CDER is to simulate the phase transition of the cloud in nature which has vapor state, liquid state and solid state, and to combine the basic ideas of evolutionary computation to realize the population evolution. The condensation growth and collision growth of cloud droplets correspond to the competitive evolution and reciprocal evolution of species in nature. Experiments on solving the function optimization problems show that this model can enhance the individual competition and survival ability, guarantee the population diversity, accelerate the convergence speed and improve the solution precision through the iterative process of competition mechanism and reciprocity mechanism. © 2012 Springer-Verlag.