制冷技术2025，Vol.45Issue(4)：46-51,84,7.DOI:10.3969/j.issn.2095-4468.2025.04.202
基于聚类与卷积神经网络的半导体工厂空调系统用能日模式分析与识别
Analysis and Recognition of Energy-Using Day Patterns of Semiconductor Factory Air Conditioning System Based on Clustering and Convolutional Neural Networks
摘要
Abstract
To study the energy consumption characteristics of air conditioning systems at different time periods,a strategy for the recognition of different energy-using day patterns of air conditioning systems is proposed in this paper.Using the K-means clustering algorithm,the operating data of a semiconductor factory air conditioning system on different dates are divided.The number of divisions is evaluated based on the elbow method and the silhouette index.Four types of different energy-using day patterns are defined,and the distribution characteristics and coefficient of performance differences under those different energy-using day patterns are analyzed.Based on this,a convolutional neural network-based pattern recognition model for energy-using days is established,and the results show that the silhouette index of the result is 0.34.The four different types of energy use day patterns have good discrepancy and representativeness,and the recognition model could recognize these four types with an accuracy of 99%.关键词
空调系统/用能模式/模式识别/K-means聚类/卷积神经网络Key words
Air conditioning system/Energy consumption pattern/Pattern recognition/K-means clustering/Convolutional neural network分类
建筑与水利引用本文复制引用
何宇轩,叶明树,张翼驰,陈焕新,程亨达..基于聚类与卷积神经网络的半导体工厂空调系统用能日模式分析与识别[J].制冷技术,2025,45(4):46-51,84,7.基金项目
国家自然科学基金(No.51876070). （No.51876070）
