基于融合聚类的冷源系统逐时能耗异常检测方法
Anomaly Detection Method of Hourly Energy Consumption for Cold Source Systems Based on Fusion Clustering
周璇 1熊智翔 1王馨瑶 1闫军威 1何敏2
作者信息
- 1. 华南理工大学机械与汽车工程学院,广州 510640;华南理工大学广州现代产业技术研究院,广州 510640
- 2. 华南理工大学广州现代产业技术研究院,广州 510640
- 折叠
摘要
冷源系统能耗异常检测是建筑空调节能的关键环节.本文提出了一种基于融合聚类的逐时能耗异常检测方法,将复杂的异常检测过程简化为规则判定问题,并以夏热冬暖地区某办公建筑冷源系统为研究对象,采用基于KMeans++聚类分析的日用能工况划分与修正方法,结合能耗特征和重要外部参数,划分出"低峰、中峰、高峰"3类日用能工况,并分析了影响工况划分的5个主要外部参数;然后,建立基于工况划分的逐时能耗基准;最后,利用该动态基准实时评价不同日用能工况下的能耗水平,并分析引起异常能耗的原因.结果表明:不同日用能工况在不同时间段的节能潜力各不相同,中峰日用能工况条件下,能耗异常发生率居中且节能潜力最大.该方法结论应用简单,适用于工程实际,并为管理者制定节能方案和提高冷源系统能效水平提供理论依据.
Abstract
Abnormal energy consumption detection of cold source systems is the key link to building air conditioning energy conservation.In this paper,an anomaly detection method of hourly energy consumption based on fusion clustering was proposed,which simplifies the complicated anomaly detection process to rule determination problem.Taking the cold source system of an office building in a hot summer and warm winter region as the research object,a condition division and correction method of daily energy consumption based on KMeans++cluster analysis was adopted to divide daily energy consumption conditions into three categories of"low peak,medium peak,and peak"based on energy consumption characteristics and important external parameters.And five main external parameters that impact the division of conditions were analyzed.Then,an hourly energy consumption benchmark based on the working condition division was established.Finally,the real-time energy consumption level of the cold source system was evaluated with an hourly energy consumption benchmark under different daily energy consumption conditions,and the causes of abnormal energy consumption was analyzed.The results show that the energy saving potential of different daily energy conditions in different time periods is different,and the incidence of abnormal energy consumption is medium with the greatest energy saving potential.This method is simple to apply,and is suitable for engineering practice.It also provides a theoretical basis for managers to make energy-saving programs and improve the energy efficiency level of cold source systems.
关键词
能耗基准/逐时能耗异常检测/日用能工况划分/融合聚类/冷源系统Key words
energy consumption benchmark/hourly energy consumption anomaly detection/daily energy consumption condition division/fusion clustering/cold source system引用本文复制引用
基金项目
广东省自然科学基金(2022A1515011128)
出版年
2024
NETL
NSTL
万方数据