摘要
智能电网系统中,费控指令通过应用服务器集群被下发到前置服务器集群,再通过前置服务器集群被下发到终端,终端接收到指令后将指令下发到特定的智能电表.在这个过程中,应用服务器集群和前置服务器集群的可用资源不对等,导致负载不均衡,影响费控服务执行的效率和成功率;前置服务器集群中前置机与终端的固定匹配方式,容易导致前置服务器集群负载不均衡,延长了费控指令完成时间,降低了费控服务的执行效率.为此深入研究费控指令的下发过程,分析存在的问题和瓶颈,提出了一种基于负载预测和负载均衡的费控服务优化调度模型.该模型首先提出一种基于时间序列的负载预测方法,实现服务器负载的预测;基于预测负载,提出一种费控指令的均衡下发算法,为不同负载的服务器下发合适规模的费控指令,以提高指令下发的效率和成功率;针对前置服务器集群可能出现的负载不均衡,提出一种基于图聚类的终端均衡布局算法.实验结果表明,本文提出的费控服务优化调度模型,均衡了两个集群和终端之间的负载,有效提高了费控指令下发的效率和成功率.
The cost control instructions are generated and processed in the application server cluster,and then are sent to the front-end server cluster.The front-end server cluster receives the cost control instructions and then sends them to the specific terminals.After receiving these cost control instructions,the terminals send them to the specific smart electricity meter.The status information of smart electricity meters could also be collected and processed in the application server cluster.In this cost control instruction delivery process,the load imbalance issue becomes the influencing factor which affects the success rate and efficiency of cost control service delivery.The load imbalance caused by available resource inequality between application server cluster and front-end server cluster could affect the cost control instruction delivery process.It could lower the execution success rate and efficiency of cost control service delivery.In addition,the stationary mapping between front-end servers and terminals is not appropriate for the dynamic cost control instruction delivery environments.It could give rise to the imbalance in front-end server cluster.The cost control instructions would be sent to the front-end server with heavy load,in which could extend the execution time of cost control service and lower the cost control instruction delivery efficiency.For the above issues in the cost control instruction delivery process,this paper deeply examines and studies the process of cost control instruction delivery,analyzes the problems and bottlenecks of the delivery process.For the load imbalance in the cost control instruction delivery process,this paper proposes an optimized scheduling model at cost control service in smart grid system based on server loads prediction and load balancing,which could alleviate the load imbalance issue and improve the execution success rate and efficiency of cost control service.In this model,a load prediction approach based on time-series data is proposed to predict the server load of application server and front-end server.The accurate prediction of server load is the basis and premise of optimized scheduling for cost control instruction delivery service.Then cost control instruction balancing sending algorithm is proposed based on the predicted server load.It could select the appropriate set of cost control instructions and send them to the appropriate server according to the predicted load.This could improve the execution success rate of cost control service and the delivery efficiency.Considering the imbalance issue in front-end server cluster,a terminal balance distribution algorithm based on graph clustering is proposed.Combining the cost control instruction history and the present cost control instructions,the associations between terminals are analyzed.These terminals are then clustered.Then the association between front-end servers and terminals are dynamically adjusted to improve the success rate and efficiency of cost control instruction delivery.Experiments demonstrate that the proposed optimized scheduling model not only make balance between sever cluster and terminals,but also improve the success rate and efficiency of cost control service.
作者
史玉良
张坤
荣以平
朱伟义
陈志勇
SHI Yu-Liang;ZHANG Kun;RONG Yi-Ping;ZHU Wei-Yi;CHEN Zhi-Yong(School of Software,Shandong University,Jinan 250101;Dareway Software Co.,ltd.,Jinan 250101;School of Information Science and Engineering,University of Jinan,Jinan 250022;State Grid Shandong Electric Power Company,Jinan 250001)
出处
《计算机学报》
EI
CSCD
北大核心
2020年第2期272-285,共14页
Chinese Journal of Computers
基金
国家重点研究发展计划项目(2018YFB1003804)
山东省泰山产业领军人才工程专项经费(tscy20150305)
山东省重点研发计划(2016ZDJS01A09)
山东省自然科学基金重大基础研究项目(ZR2017ZB0419)资助.
关键词
智能电网
智能服务
调度优化
费控指令
资源管理
smart grid
intelligent service
scheduling optimization
cost control instruction
resource management
