A Hybrid Harmony Search Approach Based on Differential Evolution

摘要

After more than ten years' development, Differential Evolution has reached an impressive state. However there are still many deficiencies, such as, its next generation just from its parents (two populations) which may restrict its diversity. Harmony Search is a new heuristic algorithm, which mimics the process of a music player to search for a perfect state of harmony in music playing. In Harmony Search, a new vector constructed from all the existing vectors rather than only from two. Harmony Search can independently consider each component variable in a vector while it generates a new vector. These features increase the flexibility of the Harmony Search algorithm and produce better solutions and overcome the disadvantage of Differential Evolution. In this paper, we propose an improved Differential Evolution method based on the Harmony Search Scheme, which we named it DEHS (Differential Evolution-harmony Search). The DEHS method has two traits. On the one hand, DEHS has the flexibility. It can adjust the values lightly in order to get a better global value for optimization. On the other hand, DEHS can greatly enhance the population's diversity. It not only uses the DE's strategies to search for global optimal results, but also utilize HS's tricks that generate a new vector by selecting the components of different vectors randomly in the harmony memory and its outside. We use this new method to solve twenty four different dimensions benchmark function optimization problems with a large number of local minima. The results show that it is significantly better than simple DE in terms of quality and stability.