Abstract: Fuzzy Adaptive Genetic Algorithm（FAGA） is a new evolutionary algorithm which applies the Fuzzy Controller（FC） to the performance and parameter control of genetic algorithm.This paper proposes an Improved Fuzzy Adaptive Genetic Algorithm（IFAGA） with two inputs and two outputs on the basis of analyzing some existed FAGAs.On the one hand，chaos initialization is adopted in order to improve the quality of initial population.On the other hand，fitness deviation is considered as one of the input parameters of FC so as to measure the discreteness of population in space.and fitness average division is considered as the other input parameter of FC so as to measure the population diversity.Through this way，the probabilities of crossover and mutation can be adaptively adjusted during the evolutionary process.The simulation results have shown that IFAGA balances the exploration and exploitation better and obtains satisfying optimization results.

Key words: Fuzzy Adaptive Genetic Algorithm（FAGA）, Fuzzy Controller（FC）, deviation average division, exploration exploitation

摘要： 模糊自适应遗传算法是将模糊控制器应用于遗传算法性能和参数控制的一种新型进化算法。提出了一种2输入和2输出的改进模糊自适应遗传算法。一方面，算法采用混沌初始化，提高了初始群体的质量；另一方面，算法将群体适应度方差作为模糊控制器的一个输入参量，来度量群体在空间分布的离散程度。将群体适应度均值商作为模糊控制器的另一个输入参量，来度量群体中个体的多样性。从而自适应地控制算法在进化过程中的交叉概率和变异概率。测试函数仿真结果表明，该算法很好地平衡了“开发”与“探测”，取得了较为满意的优化结果。

关键词: 模糊自适应遗传算法, 模糊控制器, 方差, 均值商, 开发, 探测