匹配GPU编程架构的直流故障筛选算法设计
发布时间：2018-08-09 13:04
【摘要】：首先,介绍了基于直流潮流模型的故障筛选算法;然后,基于GPU的软硬件架构特点,提出了在任务分配、PCIe总线数据交互、线程分配方式、内存访问模式4个方面GPU算法设计应遵循的通用准则;最后,对GPU加速的故障筛选算法进行了详细设计和优化,提出了一种匹配GPU编程架构的并行直流故障筛选算法。算例分析结果表明,所提算法计算9 241节点算例的绝对执行时间仅为188 ms,相对于6核心CPU算法取得了11倍加速;所提4个设计准则具有通用性,可被应用于其他电力系统算法的GPU加速。
[Abstract]:First of all, the fault screening algorithm based on DC power flow model is introduced, and then, based on the characteristics of hardware and software architecture of GPU, the data interaction and thread allocation mode of PCIe bus in task allocation are proposed. Four aspects of memory access mode GPU algorithm design should follow the general criteria. Finally, the GPU accelerated fault screening algorithm is designed and optimized in detail, and a parallel DC fault screening algorithm matching GPU programming architecture is proposed. The simulation results show that the absolute execution time of the proposed algorithm is only 188ms, which is 11 times faster than that of the 6 core CPU algorithm, and the four design criteria are universal. GPU acceleration can be applied to other power system algorithms.
【作者单位】： 华南理工大学电力学院;广东电网有限责任公司电力调度控制中心;国电南瑞科技股份有限公司;
【基金】：国家自然科学基金资助项目(51177019)~~
【分类号】：TM74
本文编号：2174159
[Abstract]:First of all, the fault screening algorithm based on DC power flow model is introduced, and then, based on the characteristics of hardware and software architecture of GPU, the data interaction and thread allocation mode of PCIe bus in task allocation are proposed. Four aspects of memory access mode GPU algorithm design should follow the general criteria. Finally, the GPU accelerated fault screening algorithm is designed and optimized in detail, and a parallel DC fault screening algorithm matching GPU programming architecture is proposed. The simulation results show that the absolute execution time of the proposed algorithm is only 188ms, which is 11 times faster than that of the 6 core CPU algorithm, and the four design criteria are universal. GPU acceleration can be applied to other power system algorithms.
【作者单位】： 华南理工大学电力学院;广东电网有限责任公司电力调度控制中心;国电南瑞科技股份有限公司;
【基金】：国家自然科学基金资助项目(51177019)~~
【分类号】：TM74
【相似文献】
相关期刊论文 前3条
1 马军海，郑万明;电阻抗层析成像（EIT）的算法设计（Ⅲ）[J];天津商学院学报;1996年01期
2 花新乐;张骞;付文杰;张旭东;王石;王清亮;栗军;;计算电缆优化分割问题的全局启发-两层求解算法[J];电网技术;2012年06期
3 ;[J];;年期
，本文编号：2174159
本文链接：https://www.wllwen.com/kejilunwen/dianlidianqilunwen/2174159.html
