基于决策树的数据中心暖通空调运行异常识别研究
Research on Abnormal Identification of HVAC Operation in Data Centers Based on Decision Tree
戴博文1
作者信息
- 1. 北京真视通科技股份有限公司,北京 100029
- 折叠
摘要
为了降低暖通空调运行异常识别结果中的噪声,引进决策树算法,以某数据中心为例,开展暖通空调运行异常识别方法的设计研究.通过传感器、仪表和监控系统等工具进行暖通空调运行工况的感知,以此为依据,实现数据中心暖通空调运行数据的采集与预处理;将不同格式、不同来源的数据进行转换和整合,实现基于决策树的元数据特征提取与类别划分;对相同类别的数据进行空间聚类处理,实现数据聚类与异常数据识别.结果表明,使用设计方法进行数据中心暖通空调运行异常识别,能将识别结果中的噪声控制在相对较低的数值.
Abstract
In order to reduce the noise in the identification results of HVAC operational anomalies,a decision tree algorithm was introduced to design and study a method for identifying HVAC operational anomalies,taking a certain data center as an example.By sensing the operating conditions of HVAC through tools such as sensors,instruments,and monitoring systems,the data center can collect and preprocess HVAC operating data based on this.Transform and integrate data from different formats and sources to achieve metadata feature extraction and category division based on decision trees.Perform spatial clustering on data of the same category to achieve data clustering and abnormal data recognition.The results show that using the designed method to identify abnormal operation of HVAC in data centers can control the noise in the identification results to a relatively low value.
关键词
决策树/异常/运行/暖通空调/数据中心Key words
decision tree/abnormal/operation/HVAC/data center引用本文复制引用
出版年
2024
NETL
NSTL
万方数据