梯级水电站群长期优化调度云计算随机动态规划算法

发布时间：2018-04-04 07:48

本文选题：随机动态规划　切入点：并行计算　出处：《中国电机工程学报》2017年12期

【摘要】：为了应对随机动态规划算法在解决梯级水电站群长期发电优化调度时的"维数灾"问题,并行化方法得到了广泛研究。单机多核并行算法扩展性不强;传统的分布式并行算法编程复杂,缺少负载均衡和容错机制。云计算平台作为新的分布式计算平台能够充分利用资源,具有诸多优势。为了探索云平台下的分布式并行随机动态规划模型,该文基于消息传递接口(message passing interface,MPI)和Spark框架分别实现了传统集群计算和云计算分布式随机动态规划算法,后者将计算模型转换为数据处理模型进行计算,并通过三库优化调度实验对算法进行了比较。算法分析及实验结果表明,基于云计算的分布式并行随机动态规划算法则可以有效利用云平台的优势,同时拥有完善的容错以及负载均衡机制,具有广阔的应用前景。
[Abstract]:In order to solve the problem of "dimensionality disaster" of cascaded hydropower station group in long-term power generation optimization, parallelization method has been widely studied in order to solve the problem of stochastic dynamic programming.The single machine multi-core parallel algorithm is not scalable and the traditional distributed parallel algorithm is complex in programming and lacks load balancing and fault-tolerant mechanism.Cloud computing platform as a new distributed computing platform can make full use of resources and has many advantages.In order to explore the distributed parallel stochastic dynamic programming model under cloud platform, this paper implements traditional cluster computing and cloud computing distributed stochastic dynamic programming algorithm based on message passing interface MPI and Spark framework, respectively.The latter transforms the computing model into a data processing model and compares the algorithm with three base optimization scheduling experiments.The algorithm analysis and experimental results show that the distributed parallel stochastic dynamic programming algorithm based on cloud computing can effectively utilize the advantages of cloud platform, and has perfect fault-tolerant and load balancing mechanism, so it has a broad application prospect.
【作者单位】：大连理工大学;
【基金】：国家自然科学基金重大计划重点支持项目(91547201)~~
【分类号】：TM73

【相似文献】