初始群体飘移遗传算法在混合整数非线性规划中的应用

本文研究了非线性混合整数规划和遗传算法各自的特点,在此基础上,提出了初始群体飘移的思想、构造了一种求解混合整数非线性规划的新的遗传算法——初始群体飘移遗传算法。数值实验表明:该算法对大范围、多峰、非光滑混合整数非线性规划问题(MINLP)有着较好的全局求解能力,在解的质量、稳定性和收敛速度等方面优于一般的演化算法。

Apply Genetic Algorithms in Mixed Integer Nonlinear Programming Based on Initial Grorp Floating

this paper discusses the features of genetic algorithms and mixed integer nonlinear programming. Based on the discussion,the definition of initial group floating and the technique of scaling integer dense grid are given.lt further proposes a new algorithm-genetic algorithms with floated intitial group which can de applied in the solution of real nonlinear programming.lt is proved by numerical experimentation that,fbr large scaling multi-apex and no smooth mixed integer nonlinear programming (MINLP),the genetic method can achieve a global solution and is superior to evolution algorithms in feasibility,stabilization and convergent speed of the solution.