Clouding algorithm: A novel multi-population evolution model and its applying to global numerical optimization

This paper proposes a multi-population evolution model, which is distinguished from other models by declaring that the fittest individuals should produce a population with many offspring rather than only one child. Therefore, special attention is spent upon evolution of populations in a community rather than individuals in a population. To model this idea, an uncertainty model, namely cloud model, is employed. By means of model a population with a cloud, inheritance and mutation of population is well unified in a natural way. Further more, we propose a novel evolutionary algorithm, namely clouding algorithm, which describes the framework of algorithms using multi-population evolution to solve global numerical optimization problems. It has no crossover operator, but significantly focuses on mutation control. Besides, a concrete clouding algorithm or the basic clouding algorithm is presented in detail. Experiments are also conducted and results demonstrate good performance of BCA when compared with other algorithms.