中文摘要：

任务调度是集群系统的关键技术之一，针对总线结构 DAG模型下的任务调度问题，提出一种基于混沌改进的遗传模拟退火算法，简称Chaos-GSA。该算法在原有遗传模拟退火算法的基础上引入混沌系统，改进种群初始化方法和交叉、变异算子，提高算法的收敛速度，并在降温时考虑染色体资源平均利用率，使具有较大资源利用率的个体更容易被选择。实验结果表明：该算法与传统 GSA算法相比，有明显的优越性，可以减少时间跨度，提高资源的利用率。

英文摘要：

Task scheduling is one of the key technologies in the cluster system.For the task scheduling problem in DAG model of bus structure,an improved genetic simulated annealing algorithm based on Chaos （Chaos-SA）was put forward.As an extension of the existing genetic simulated annealing algorithm,the Chaos-GSA algorithm embedded chaotic systems to create better initial data, refined crossover and mutation operators,yielding improved convergence rate.In the process of the annealing,the factor of the average rate of resources was fully considered,so that the larger rate of resources of individuals,the more likely they were selected.The simulation results showed that the algorithm,compared with the traditional GSA,had obvious advantages of reducing the makespan and improving the usage of resources.