基于分布估计算法的Web服务组合优化

摘要

分布估计算法( EDAs)将遗传算法和统计学习相结合，并利用概率模型来描述变量之间的相互关系，提高解决高维问题的效率，降低时间复杂性，最终求出最优解。文中将分布估计算法应用于解决Web服务组合问题，并提出了基于分布估计算法的Web服务组合优化模型。仿真实验采用了EDAs中的基于群体的增量学习算法( PBIL)，分析了服务类的数量以及采用精英保留策略对优化结果的影响。结果表明采用了精英保留策略的分布估计算法求解Web服务组合的问题是可靠有效的。%Estimation of Distribution Algorithms ( EDAs) combines genetic algorithm and statistical learning,and uses probability models to describe the relationships between variables,which improves the efficiency of solving the high-dimensional problem,and reduces the time complexity,thus achieving the optimal solution. In this paper,EDAs is applied to Web Services composition problem and present a Web Services composition optimization model based on EDAs. Simulation experiments use Groups-Based Incremental Learning ( PBIL) algorithm in EDAs,and analyze the impact of number of service classes and elitist policies on optimization results. Experiments results show that EDAs with elitist strategy for solving Web Services composition problem is reliable and efficient.