差分进化算法DE(DifferentialEvolution)具有记忆个体最优解和种群内部信息共享的特点,是一种基于实数编码的、具有保优思想的简单、高效的新型进化算法.针对云计算任务调度问题,提出了一种考虑时间和成本约束的基于改进DE的调度算法(TcDE).该算法采用随机初始化种群,合法化修复变异个体,并采用贪婪直接交叉,在变异操作之后,加入新的选择机制,防止交叉操作破坏变异出的优良个体.在仿真实验中分别进行了TCDE与只考虑时间的TDE ,只考虑成本的CDE的功能验证实验,TDE与遗传算法的性能对比实验.实验结果证明:改进后的差分进化算法能够适用于云计算环境下的任务调度问题,并取得良好的调度效果.
Differential evolution(DE), with the characteristics of remembering individual optimal solution and information sharing, is a real coded and excellent security and simple efficient new kind of evolutionary algorithm. To solve the problem of task scheduling for cloud computing, a task scheduling algorithm is pro posed considering the time and cost constraints based on improved differential evolutionary(TCDE). Ran dom algorithm is used to generate the initial population, legalized method is used to repair mutation and greedy crossover is used, too. At the same time, after the mutation operator, a new selection mechanism is added to prevent the destruction of fine individual, which is produced by mutation from crossover opera tion. The results of functional verification test of TCDE,TDE(only consider the time)and CDE(only con sider the cost), and performance analysis test of TDE and GA show that the TCDE can be applicable to task scheduling problem in the cloud computing environment and achieve good scheduling.