分布估计算法在多维函数优化中的应用  

Application of Estimation of Distribution Algorithm in Optimizing Multi-dimensional Function Model

作　　者：谷月霞[1] 

机构地区：[1]聊城职业技术学院计算机系,山东聊城252000

出　　处：《武汉职业技术学院学报》2010年第1期54-56,共3页Journal of Wuhan Polytechnic

摘　　要：分布估计算法是一类新的进化算法,它通过统计在当前群体中选出的个体信息给出下一代个体分布的概率统计,用随机取样的方法生成下一代群体。文章将建立在一般结构Gauss网络上的分布估计算法应用于多维函数的优化并与传统的粒子群优化算法进行了比较。仿真实验结果表明,分布估计算法可以在很短的时间内收敛至全局最优解,避免了传统算法的不足,提高了函数优化的准确性。Estimation of distribution algorithm （EDA） is a new kind of evolutionary algorithm. The probability estimation of the individual distribution in next generation is made by calculating the data of selected individual in current generation, and the next generation of group is formed by random sampling. Here, the estimation of distribution algorithm based on generic Gaussian Networks is applied to optimizing the Multi-dimensional Functions, and is compared with traditional Particle Swarm Optimization （PSO）. The simulated results shows that EDA applied to optimizing the Multi-dimensional Functions model can converge on the global optima in very short time, avoiding the defects of the traditional algorithm and enhancing the accuracy.

关 键 词：分布估计算法 PSO算法 多维函数优化 

分 类 号：O24[理学—计算数学]