论文部分内容阅读
本文在分析标准遗传算法的优点和不足的基础上,基于AER模型提出了一种新的遗传算法——Multi-A-gent遗传算法.它利用Agent的局部感知、竞争协同和自学习等特性来实现生物对环境的自适应,从而实现全局优化计算.理论分析证明这种算法是以概率1收敛的.在实验中,我们首先用10个维数为30的标准测试函数来全面测试算法的性能,然后用50～200维的Rastrigin函数来测试算法处理高维函数的能力.结果表明本文算法具有较强的全局优化能力,鲁棒性强,且具有良好的处理高维函数的能力.
Based on the analysis of the advantages and disadvantages of standard genetic algorithms, this paper proposes a new genetic algorithm based on the AER model, Multi-A-gent genetic algorithm, which takes advantage of the local perception, competitive coordination and self-learning of Agent To achieve the biological adaptation to the environment in order to achieve the global optimization calculation.The theoretical analysis proves that this algorithm converges with probability 1. In the experiment, we first use 10 standard test function of dimension 30 to fully test the performance of the algorithm , And then use Rastrigin function of 50 ~ 200 dimension to test the ability of the algorithm to deal with high dimensional functions.The results show that the proposed algorithm has strong global optimization ability, strong robustness and good ability to deal with high dimensional functions.