您当前访问的的浏览器版本过低,为了给您带来更好的体验,建议您升级至Edge浏览器或者推荐使用Google浏览器
取消

基于AI的运营级IDC节能研究

发布时间:2020-10-22  作者:曾宇, 袁祥枫, 王海宁  阅读量:

基于AI的运营级IDC节能研究

曾宇1,袁祥枫1,王海宁2
(1. 中国电信AI研发中心,中国 北京 102209;2. 英特尔(中国)有限公司,中国 北京 100013 )

摘要:通过分析运营商互联网数据中心(IDC)机房动环、空调、机柜微环境等数据,得到机房画像,并对机房相关参数之间的映射关系进行AI建模。再利用得到的模型,以及数据间相关性,通过机房历史功耗数据,可以对机房未来功耗趋势进行预测,从而找到机房能耗优化依据。通过引入模糊控制模型,机房运维的人工控制经验可以得到固化,形成节能控制策略规则库。通过在试点省份验证,控制策略可普遍用于同类型空调的应用场景。与传统IDC节能方法比较,提出的方法可结合机房能耗特征,可实现“千房千面”,且节能成效显著。
关键词:IDC智慧节能;机房画像;能耗管理


Carrier Grade IDC Energy Saving Research Based on AI

ZENG Yu1,YUAN Xiangfeng1,WANG Haining2
(1. China Telecom Beijing Research Institute, Beijing 102209, China; 2. Intel China Ltd, Beijing 100013, China )

Abstract: Through the analysis of the operator Internet data center (IDC) dynamic environment, air conditioning, cabinet microenvironment data, the IDC portrait is obtained, and artificial intelligence (AI) models can be derived by mapping relevant parameters of the IDC. By using the obtained model and the data correlation, the IDC power consumption trend can be predicted through the historical power consumption data of IDC. This will form the basis for energy optimization of the IDC. By introducing the fuzzy control model, the manual control experience of IDC operation and maintenance can be solidified to form the rule base of energy-saving control strategy. Through verification in pilot provinces, the control strategy can be generally used in the same type of cooling application scenarios. Compared with the traditional IDC energy-saving method, the algorism and method proposed in this paper can combine the characteristics of the energy consumption of the IDC to achieve “tailor made” energy solutions, and the results is beyond expectation.
Keywords: AI based IDC energy saving; IDC profile; energy management

分享到: