期刊文献+

基于机器学习的电网信息通信服务器智能优化 被引量:9

Intelligent Optimization of Information Communication Server of Power Grid Based on Machine Learning
下载PDF
导出
摘要 随着中国电网向智能化、网络化、自动化发展,电网信息通信服务器承载着电网信息网络信息传输中的核心业务。信通服务器一般采用线程池技术来应对逐渐增多的用户请求,而选择合适的线程池尺寸成为了决定服务器性能的关键因素。提出一种基于支持向量机的信通服务器动态线程池智能优化模型,来动态减少用户的响应时间。首先,通过大量的信通服务器性能实验数据构造原始训练样本集,然后经过改进的流体优化算法搜索支持向量机的最优超参数,最后通过训练好的支持向量机预测不同电网用户场景下的最优线程池尺寸,从而实现对信通服务器的智能优化。通过辽宁省电网信通服务器的实验表明,基于改进流体优化算法的支持向量机智能线程池技术获得了更高的预测精度,减少了服务器的用户响应时间。 With China's power grid becoming more intelligent,networked and automatized,the information communication server(ICS)plays a key role in the information transmission of power grid.Generally,the thread pool technology is used in ICS to deal with the increasing users’requests,and the key of server performance is how to choose the appropriate size of the thread pool.So an intelligent optimization model of dynamic thread pool in ICS based on support vector machine(SVM)was proposed to reduce the response time.Firstly,the original training set was made from a large amount of experimental data of the performance of ICS.Then,the fluid optimization algorithm(FSO)was improved to search the optimal hyper parameters of SVM.Finally,the trained SVM was used to predict the best size of thread pool in different user scenarios,so that the optimization of ICS was accomplished intelligently.The experiments on ICS of Liaoning Power Grid show that the intelligent thread pool technology based on SVM and IFSO achieves higher prediction accuracy and reduces the response time of the server.
作者 申扬 于海 尹晓华 SHEN Yang;YU Hai;YIN Xiao-hua(State Grid Liaoning Electric Power Supply Co., Ltd., Information & Telecommunication Branch, Shenyang 110006, China)
出处 《科学技术与工程》 北大核心 2020年第32期13302-13308,共7页 Science Technology and Engineering
基金 国家自然科学基金(51437003)。
关键词 服务器智能优化 支持向量机 流体搜索优化算法 信息通信服务器 线程池调优 intelligent optimization of server support vector machine fluid search optimization algorithm information communication server thread pool optimization
  • 相关文献

参考文献10

二级参考文献58

共引文献48

同被引文献76

引证文献9

二级引证文献48

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部