基于GA的控制系统中实时任务的优化调度  

Optimal Scheduling for Real-Time Tasks in Control Systems Based On GA

作　　者：刘怀[1] 谢东风[2] 刘宁[3] 黄建新 

机构地区：[1]南京师范大学电气与自动化工程学院,江苏南京210042 [2]云南省烟草公司,云南昆明650011 [3]东南大学自动化研究所,江苏南京210096

出　　处：《南京师范大学学报（工程技术版）》2007年第4期13-17,共5页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：南京师范大学青年科学基金(200411XQNBDQ41)资助项目

摘　　要：控制系统中任务的调度不仅影响系统的资源利用率,而且影响系统的控制性能,是控制系统设计和实现的关键技术之一.首先分析了周期性实时任务,并给出了任务模型.基于此,采用了EDF调度算法,给出了任务的可调度性条件.根据系统的性能指标与任务的采样周期和控制延迟之间的关系,对系统性能进行优化.由于采样周期和控制延迟之间存在相互影响,为此采用了浮点数编码的遗传算法对采样周期进行优化,以提高系统的性能.为了加快收敛速度和不陷入局部极值,采用了排序选择算法、算术交叉算子和非一致变异算子.仿真结果表明,通过采用遗传算法,在保证系统中任务实时性的条件下,可以显著地提高系统的性能指标.Task scheduling in the control system influences not only the resource utilization of the system, but also the control performance of the system, and thus is one of the key techniques for designing and realizing the system. Firstly, real -time periodic task is investigated and task model is given. Based on these, EDF algorithm is adopted to schedule this kind of task and schedulable condition is prevented. According to the relationship of the performance of control system and sampling periods and control delays of tasks, the performance is optimized. Because sampling periods and control delays have influences on each other, genetic algorithm based on floating-point coding is adopted to optimize sampling period so as to improve the performance of the system. In order to converge to global optimum more quickly and not to plunge a local extremum, rank-based selection algorithm and arithmetic crossover operator and nonuniform mutation operator are adopted. Simulation results indicate that the performance index of control system can be improved obviously by adopting genetic algorithm under the condition of guaranteeing the real-time of the tasks in the system.

关 键 词：控制系统 调度算法 遗传算法 采样周期 控制延迟 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]