Genetic algorithm (GA) is a suitable algorithm prototype for numerical optimization.Based on a three-parent crossover (TPC) and a diversity operator,the performance of GA is greatly improved,though it is limited by a few algorithmic parameters.The adaptation of algorithmic parameters in TPC and diversity operator is investigated.The key parameters are generated based on normal distribution in each iteration.Experiments are conducted on a set of 13 mathematical functions.The results of the algorithm with and without parameter adaptation are compared fromf1 tof13.Furthermore,the convergence processes off4 and f10 are presented which show that the adaptive GA-TPC is more efficient and more robust than the GA-TPC algorithm.%遗传算法(GA)是一种适合于数值优化的算法原型,基于1个三父体交叉(TPC)和1个多样性算子虽然可使GA的性能得到很大改进,但仍受制于几个算法参数.在此基础上,对TPC和多样性算子中算法参数的自适应遗传算法进行研究.算法的关键参数在每次迭代中由正态分布生成,并在1组13个数学函数集上施行.对原算法与添加参数适应算法的结果在函数f1～f13上进行对比,并给出了f4和f10的收敛过程,分析表明自适应GA-TPC算法比原算法在解决具体问题时更加高效和稳定.
展开▼
