高速收敛混沌粒子群算法的云计算任务调度

Cloud Computing Task Scheduling of High-Speed Convergence of Chaotic Particle Swarm Optimization

针对传统粒子群算法在处理云计算任务调度问题时,存在求解精度不高、容易陷入早熟收敛等缺陷,提出一种改进的高速收敛混沌粒子群算法.首先,采用混沌序列对初始化过程进行优化;其次,利用适应度方差对早熟现象进行有效诊断,并对算法在负梯度方向进行修正,使其跳出局部最优,实现高速收敛.仿真实验表明:改进后的粒子群算法能有效地避免早熟,收敛速度及求解精度都明显提高,非常适合云计算任务调度.

In this paper, we ＇proposed an advanced high speed of convergence chaotic particle swarm algorithm to adjust the common problems of traditional particle swarm algorithm such as low accuracy and easily trapped in premature conver- gence during the cloud computing task scheduling. Firstly, the initial process was optimzed by chaotic sequence. Then, the effective diagnosis of premature phenomenon was determined by fitness variance. The algorithm correction was per- formed by negative gradient direction, which could jump out the local optimum and achieve high speed of convergence. Simulation experiments show that the improved particle swarm algorithm can effectively avoid premature, enhance conver- gence speed and solution accuracy, which is suit;able for cloud computing task scheduling.