基于种族优生的进化规划用于混合非线性整数规划

Solving Mixed Integer Non-Linear Programming Problems by the Evolutionary Programming Based on the Prepotency of Races

混合非线性整数规划是在许多工程应用中经常遇到的重要问题。本文提出一种基于种族优生的进化规划算法用于求解混合非线性整数规划。一方面,该算法基于多种群并且每一代都选择各种群的最优秀个体作为下一代的种群祖先。另一方面,该算法的进化步长、种群规模和处理约束条件时所取的参数在进化过程中是动态变化的。实验表明该方法求解混合非线性整数规划问题的仿真结果优于现有的研究成果(GA,ES,SA)。

The global optimization of mixed integer non-linear problems (MINLP) constitutes a major area of research in many engineering applications .In this paper, a novel evolutionary programming is proposed as a valid approach to the optimization of mixed integer non-linear problems. On one hand, the method is based on multiple groups and the best individual of very group in the present generation is selected as the ancestor of the next generation. On the other hand, the evolutionary step, the size of the group and the parameter about handling constraint are dynamically altered in the evolutionary process. Computer simulation results show it excels the research fruit of Genetic Algorithm (GA), Evolution Strategic (ES) and Simulated Annealing (SA).