基于分布式系统的资源运维管理可靠性提升方法  

Reliability improvement of resource operation and maintenance management based on distributed system

作　　者：孙乔 付兰梅 裴旭斌[2] 孙甲松[3] 

机构地区：[1]北京国电通网络技术有限公司,北京100070 [2]国网浙江省电力公司,杭州310007 [3]清华大学电子工程系,北京100084

出　　处：《计算机应用》2017年第A01期243-245,254,共4页journal of Computer Applications

基　　金：国家电网公司科技项目(SGZJ0000BGJS1500433);国网信息通信产业集团有限公司科技项目(SGITG-KJ-JSKF[2015]0003)

摘　　要：提出一种基于分布式网络系统资源运维管理的可靠性提升方案,采用深度神经网络学习算法进行基于分布式网络系统的判断调度决策估计方法。通过这种资源分配方案,可以满足电网信息系统对一致性和监测评估处理方面的要求。在内网环境中的评估实验表明,基于本方案的资源运维管理方式,其使用率和成功执行率在任务个数较多时仍能维持较好的比率,且其任务管理策略也优于其他方式。在较难预测的复杂场景下,在维持计算复杂度的条件下对全网络系统的资源运维管理起到优化作用,同时在此过程中,对性能和相应速度等指标的波动具有快速处理能力。总之,这种方法不仅能提高分布式网络系统的使用率和成功执行率,还具有可靠性、高精度及可扩展性上的优势。This paper proposed a reliable distributed network system for resources maintenance and management based on lifting scheme by the deep neural network learning algorithm to carry out the estimation judgement, scheduling and decision based on distributed network system. This resource allocation scheme can meet the requirements of the power network information system to consistency and the monitoring and evaluation process. Intranet environment evaluation experiments show that by using the maintenance resource management based on the proposed scheme, the use rate and success rate can still maintain good percentage when the number of tasks is large and the task management strategy is better than the other ones.The resource operation and maintenance management of the whole network system is optimized while maintaining the computational complexity. At the same time, the volatility of the corresponding speed and other indicators was rapidly processed. This method not only can improve the utilization rate and successful execution rate of the distributed network system, but also has the advantages of reliability, high precision and expansibility.

关 键 词：资源运维管理 分布式网络系统 深度神经网络 优化设计 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]