摘要
传统遗传算法在种群初始化的时候,普遍采用均匀取种法或随机取种法,这些方法生成的种群的平均适应度比较低,难以保证算法的搜索效率。文中提出一种改进的遗传算法用于QoS敏感的Web服务组合,采用两种不同的算法进行服务选择,避免了随机生成初始种群给算法带来的负面影响。并且,该算法将路径模板化以减少服务组合的工作量,用染色体可变长的编码方式来解决组合服务的多路径选择问题。通过仿真实验,与传统的算法相比,所提出的算法在实现服务组合时收敛更快,最优解的适应度更高。
Even or random selecting is the common method used for generating initial population in genetic algorithm,however,the aver- age fitness of the population generated by this method is low, and it is hard to ensure the .searching efficiency of algorithm. In this study, propose a novel genetic algorithm （GA） for handling QoS-aware Web service composition, combining two initialized algorithms with GA at initialization stage to improve the algorithm effectiveness. Besides, build a path-template and variable length chromosomes service composition solution, for template paths will make the work easy and variable length chromosomes can support multi-path QoS-aware service composition. The superiority of the algorithm is analyzed theoretically and its effectiveness is demonstrated by experimental re- suits.
出处
《计算机技术与发展》
2012年第8期89-92,共4页
Computer Technology and Development
基金
国家科技支撑计划(2007BAH17B04)
