电力云计算任务调度系统研究

发布时间：2018-09-17 15:20

【摘要】：随着我国智能电网技术不断地发展和深化，电力信息系统面临着数据海量、分布异构、处理复杂、使用繁琐、维护困难等问题，而云计算技术具有强大处理能力、动态性、灵活性、虚拟化以及面向服务等特性，正是一种应对上述问题的有效解决方案。因此可以将云计算引入到电力系统中，构建电力系统的专属云—电力云。在电力云中，，任务调度系统是一个重要的组成部分，是提高电力云整体并发性、保证用户任务和电力云资源合理分配、提高电力云的计算性能的关键。因此本文对电力云任务调度系统进行研究，主要从调度系统模型以及调度算法两个方面进行了研究。在对云计算基础理论和电力信息平台研究的基础上，针对复杂的用户任务，提出了一种基于改进MapReduce模型的任务调度系统，保留了原有MapReduce模型的数据并行性优点的同时，实现在架构上进行任务分解和并行计算，大大扩展了MapReduce模型的并行性。在该任务调度系统模型的基础之上，针对不同用户任务的需求，提出了一种电力云任务调度算法，该算法以遗传算法为原型，通过调整算法的多维约束条件来满足不同用户的需求。最后本文提出来的任务调度算法与多种其他的算法进行仿真实验对比，实验结果表明该算法是一种十分有效的任务调度算法。
[Abstract]:With the continuous development and deepening of smart grid technology in China, power information system is faced with the problems of massive data, heterogeneous distribution, complex processing, cumbersome use, difficult maintenance and so on. Cloud computing technology has a strong processing ability, dynamic, and so on. Flexibility, virtualization, and service-oriented features are an effective solution to these problems. Therefore, cloud computing can be introduced into the power system to build the power system's exclusive cloud-power cloud. Task scheduling system is an important part of power cloud, which is the key to improve the concurrency of power cloud, ensure the rational distribution of user's tasks and power cloud resources, and improve the computing performance of power cloud. Therefore, this paper studies the power cloud task scheduling system, mainly from two aspects: the dispatching system model and the scheduling algorithm. Based on the research of cloud computing theory and power information platform, a task scheduling system based on improved MapReduce model is proposed for complex user tasks, which retains the advantages of data parallelism of the original MapReduce model. The implementation of task decomposition and parallel computing in architecture greatly extends the parallelism of MapReduce model. Based on the model of the task scheduling system and according to the needs of different users, a power cloud task scheduling algorithm is proposed, which is based on genetic algorithm (GA). The multi-dimensional constraints of the algorithm are adjusted to meet the needs of different users. Finally, the task scheduling algorithm proposed in this paper is compared with many other algorithms. The experimental results show that the algorithm is a very effective task scheduling algorithm.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP301.6;TM73

【参考文献】