Differential evolution algorithm was introduced by R.Storn and K.Price in 1995. And it has already been applied successively to many areas such as multimodal function optimization, neural network learning, digital filter design, multi-objective optimization and so on.In this paper, it brings into focus on differential evolution's representation in function optimization. Four novel schemes based on differential evolution are proposed. And two results about on differential evolution are theoretically proved. The summary is following:1. An modified differential evolution algorithm is presented. In the proposed algorithm, the difference vector which is employed in mutating each individual by simple differential evolution algorithm is recorded. Each mutated individual is taken as the center of a neighborhood, the corresponding difference vector as radius. Then search in the neighborhood again. The modification increases the local searching ability of differential evolution algorithm.2. Hybrid differential evolution algorithm based on simulated annealing is designed and tested by several nonlinear function optimization problems. The results indicated the proposed algorithm can improve the efficiency of differential evolution algorithm and much more robust than simply differential evolution.3. Coupling differential evolution and simulated annealing algorithm based on niche is proposed. The results show that the proposed algorithm can find all global optimum points quickly without strict request for parameters. Niche can improve and preserve the population diversity.4. A new scheme based on Pascal distribution is proposed. And it is proved theoretically that the proposed scheme is much more effective than the primary algorithm.5. A new heuristic scheme based on an eclectic differential evolution... |