【摘要】 目前对混合整数非线性规划问题做的工作甚少 ,这里在研究了混合整数非线性规划和遗传算法各自特点的基础上 ,提出初始群体飘移思想和变尺度整型细密网格技术 ,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。经理论分析和数值试验表明 :该算法对大范围、多峰、非光滑非线性规划问题有较好的全局求解能力 ,在解的精度、稳定性和收敛速度方面均优于一般的算法。更多还原

【Abstract】 Until now the research for mixed integer nonlinear programming has hardly made a ny progress. In the paper, genetic algorithms and mixed integer nonlinear progra mming are first discussed. Then based upon the definitions of initial group floa ting and techniques of scaling integer dense grid, a new algorithm with initial group floating is presented for mixed integer nonlinear programming. Theoretical analysis and numerical tests show that the genetic method can perform a good gl obal optimum solution for large scaling multi-apex and non smooth mixed integer nonlinear programming (MINLP) with better results than other algorithms used to resolve MINLP in the feasibility, stabilization and convergent speed.更多还原