基于杂交遗传算法的多处理器硬实时容错调度算法

发布时间：2018-07-07 21:50

本文选题：遗传算法 + 硬实时系统　；参考：《计算机应用研究》2016年09期

【摘要】：传统的硬实时容错调度算法获得了较好的容错性能,但其任务拒绝率、处理器分配偏差比例以及最早完成时间等性能参数不佳,对此提出了一种基于杂交遗传算法的优化方案,并对传统的硬实时容错算法进行优化。采用了中心型调度模型,并采用了任务备份方案来实现容错能力。将任务拒绝率、处理器分配偏差比例以及最早完成时间三个优化参数作为遗传算法适应度目标函数的三个带权分量,对其进行优化,通过遗传算法的杂交与迭代计算获得了优化的结果。最终使用不同的任务数量与处理器数量的组合对本算法与传统算法进行对比实验,结果可看出本算法的三个优化参数明显优于传统算法,且总适应度值亦比传统算法有明显改进。
[Abstract]:The traditional hard real-time fault-tolerant scheduling algorithm has better fault-tolerant performance, but its performance parameters such as task rejection rate, processor allocation deviation ratio and earliest completion time are not good. Therefore, an optimization scheme based on hybrid genetic algorithm is proposed. And the traditional hard real-time fault-tolerant algorithm is optimized. The central scheduling model is adopted, and the task backup scheme is adopted to realize the fault-tolerant ability. Three optimization parameters, task rejection rate, processor allocation deviation ratio and earliest completion time, are considered as three weighted components of the fitness objective function of genetic algorithm. The optimization results are obtained by hybrid and iterative calculation of genetic algorithm. Finally, the comparison experiment between this algorithm and the traditional algorithm is carried out using the combination of different task number and processor number. The results show that the three optimization parameters of this algorithm are obviously superior to the traditional algorithm. And the total fitness value is obviously improved compared with the traditional algorithm.
【作者单位】：顺德职业技术学院电子与信息工程学院;华南理工大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(51175183) 佛山市产学研专项资金项目及顺德职业技术学院资助项目(2012HC100303) 广东省教育科研“十二五”规划研究项目(2012JK305)
【分类号】：TP332

【相似文献】