A hybrid constraints scattered genetic algorithm with interior point method
Wenxing Xu, Zhenyu Wang, Qunxiong Zhu, Zhiqiang Geng · 2011
While using genetic algorithm to solve constrained optimization problems, some of its shortcomings appear such as difficulty in obtaining feasible individuals in many strong constraint conditions and poor local search ability. In this paper, an algorithmic thought of constraints scattering is presented to divide the complete constraints of a problem into several sub-populations for processing, which effectively improves the global search ability of the new algorithm. Experiments show that with this operation population can achieve better performance, which means containing more feasible solutions with larger feasible solution diversity, can be generated during the iteration of genetic algorithm. Then further combined with the interior point method, a new hybrid constraints scattered genetic algorithm with interior point method (CSGA-I) is proposed to ensure the local search ability while searching the whole feasible region. Experiments and comparisons over a set of standard test functions demonstrate that our approach has a better solution precision at less computation cost than most of the other algorithm reported in literature.