水电系统中长期发电调度多核并行逐步优化方法  

Multi-core parallel progressive optimization algorithm for mid- / long-term generation dispatch of hydropower system

作　　者：冯仲恺[1] 牛文静[1] 廖胜利[1] 程春田[1] 苏华英[2] 汪明清[2] 唐建兴[2] FENG Zhongkai NIU Wenjing LIAO Shengli CHENG Chuntian SU Huaying WANG Mingqing TANG Jianxing(Institute of Hydropower ＆ Hydro Information, Dalian University of Technology,Dalian 116024,China Guizhou Electric Power Dispatching and Communication Bureau,Guiyang 550000,China)

机构地区：[1]大连理工大学水电与水信息研究所,辽宁大连116024 [2]贵州电网公司电力调度控制中心,贵州贵阳550000

出　　处：《电力自动化设备》2016年第11期75-81,共7页Electric Power Automation Equipment

基　　金：国家自然科学基金重大国际合作项目(51210014);国家自然科学基金重大计划重点支持项目(91547201);国家重点基础研究发展计划(973计划)资助项目(2013CB035906)~~

摘　　要：随着智能电网运行与管理的科学性和时效性要求不断提升,迫切需要高效方法保证水电优化调度计算的求解精度与运算效率。在逐步优化算法并行性分析的基础上,结合Fork/Join框架提出多核并行逐步优化算法。该方法将多阶段问题分解为若干两阶段子问题,在求解各子问题时采用分治策略结合多核并行计算技术进行寻优,以提高运算效率。某水电站群调度实践结果表明,所提方法可在保证求解质量的同时,充分利用并行计算资源大幅缩短运算耗时,且计算规模越大,性能优势越显著。High efficient method is urgently needed to guarantee the solution accuracy and computation efficiency of hydropower optimization dispatch calculation for the scientific and effective operation and management of smart grid. The parallelism of POA（Progressive Optimality Algorithm） is analyzed and a multicore parallel POA is proposed based on the Fork / Join framework,which divides a multistage problem into several dual-stage sub-problems and combines the divide-and-conquer strategy with the multi-core progressive optimization technique to search the optimal solution of sub-problem for improving the computation efficiency.Its application in the dispatch of a hydropower station group shows that,the computation efficiency is greatly improved while the solution quality is guaranteed;the larger the computation scale is,the more prominent its performance advantage becomes.

关 键 词：多核并行 优化 水电 发电 调度 降维 

分 类 号：TM862[电气工程—高电压与绝缘技术]