基于拉丁超立方体抽样和免疫机制的改进遗传算法

Improved genetic algorithm based on Latin hypercube sampling and immune mechanism

针对遗传算法求解问题中保持群体多样性能力不足、早熟以及求解成功率低等缺点,依据拉丁超立方体抽样方法对遗传算法中的交叉算子进行重新设计;结合免疫机制定义染色体浓度、提供选择依据,提出了一种新遗传算法。利用旅行商问题以及最大子团问题为实例对新算法进行了验证,实验结果表明新算法在解的质量、收敛速度等各项指标上均好于经典遗传算法和佳点集遗传算法,说明了新算法的优越性与可行性。

Concerning the defects of Genetic Algorithm（GA） in the deficiency of keeping population diversity,prematurity,low success rate and so on,the crossover operation in GA was redesigned by Latin hypercube sampling.Combined with immune mechanism,chromosome concentration was defined and selection strategy was designed,thus an improved genetic algorithm was given based on Latin hypercube sampling and immune mechanism.The Traveling Salesman Problem（TSP） and the Maximum Clique Problem（MCP） were used to verify the new algorithm.The results show,in terms of solution quality,convergence speed,and other indicators,the new algorithm is better than the classical genetic algorithm and good-point-set genetic algorithm.