基于空调使用行为记录的用户热偏好预测模型及控制方法分析  

Analysis of User Thermal Preference Prediction Model and Control Method Based on Air Conditioning Usage Behavior Records

作　　者：张静思 ZHANG Jingsi(Midea Group Central Research Institute,Shanghai 201702,China)

机构地区：[1]美的集团中央研究院,上海201702

出　　处：《电子技术（上海）》2024年第3期164-167,共4页Electronic Technology

摘　　要：阐述预测空调用户的个体热偏好有助于智能空调控制技术的发展。分析空调中的嵌入传感器数据以及用户与这些设备的交互行为,采用机器学习算法对空调的设定温度变化进行预测。结果表明,不同用户的空调使用行为在设定温度偏好和调节时间上存在较大差异。超过60%的用户倾向于将空调温度设置在25~28℃范围内。机器学习模型预测升高设定温度和降低设定温度的准确率为72.1%~87.3%,通过添加月和小时作为输入特征,模型的性能随着样本的增大而提高。This paper expounds that predicting individual thermal preferences of air conditioning users can contribute to the development of intelligent air conditioning control technology.It analyzes the embedded sensor data in the air conditioner and the interaction behavior between users and these devices,and uses machine learning algorithms to predict the set temperature changes of the air conditioner.The results indicate that there are significant differences in temperature preferences and adjustment time among different users in their air conditioning usage behavior.More than 60%of users tend to set the air conditioning temperature within the range of 25-28℃.The accuracy of machine learning models in predicting the increase and decrease of set temperatures is 72.1%~87.3%.By adding months and hours as input features,the performance of the model improves with the increase of samples.

关 键 词：数据驱动模型 智能家居 行为分析 空调控制 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TS941[自动化与计算机技术—控制科学与工程]