Abstract:

In order to effectively apply fruit fly optimization algorithm in multimodal function optimization, we propose an improved fruit fly optimization algorithm which optimizes multimodal function―a mixed fruit fly optimization algorithm based on the good point set and niche technology. Firstly, we employ the concept of good point set to construct the initial population and distribute it more evenly in the feasible region, which can produce a better pattern diversity than random distribution, improve the search ability, efficiency and stability of the algorithm. Secondly, we adopt the niche technology to improve the algorithm's search mode and better maintain the diversity of the population, thus enabling the population to locate peaks quickly. The niche entropy can quantify the diversity of the population and choose the direction of evolution. When the niche entropy is lower than the set threshold, together with the good point searching, a new group is generated, which provides perturbation and maintains population diversity. Otherwise, searches for each peak are finely done. We carry out two kinds of simulations on 7 test functions, and the results show that the proposed algorithm cannot only find the global extreme values with high efficiency and accuracy, but also locate all the global extreme values and multiple sub optimal extremum at a higher precision, and it shows strong multipeak search ability.

Key words: fruit fly optimization algorithm, multimodal function optimization, good point set, niche technology, niche entropy