Abstract: The paper proposes a new optimization hybrid algerithm based on PSO and GA.It divides the samples into group N，and then each group operates by PSO or GA of different parameters.It selects the optimaler number as a global optimum at every circulation，which makes its result be better than both PSO and GA，then improves the overall performance of the algorithm.Compared with other hybrid optimization algorithm，it does not undermine the independence of PSO and GA，but combins the two algorithm by the global optimums only.The experimental results show this approach has shown better convergence and stability.

Key words: particle swarm, genetic algorithm, optimization, hybrid algorithm

摘要： 提出了一种基于粒子群和遗传算法的新混合算法。该算法首先将样本集分为N组，每一组分别进行不同参数的粒子群或遗传运算，在每一步的迭代中选取了粒子群算法和遗传算法的最优值作为全局最优，使每一步的迭代都优于单一的PSO和GA算法，进而提高了算法整体的性能。与其他混合最优化算法不同的是，该算法没有破坏粒子群和遗传算法的独立性，而是仅通过全局最优样本把两个算法结合在一起。在经典测试函数的仿真实验中，新算法表现了更好的寻优性能及寻优稳定性。

关键词: 粒子群, 遗传算法, 函数优化, 混合算法