基于LBD-WA反馈提升的空调智能控制方法研究  

Research on Intelligent Control Method for Air Conditioning Based on LBD-WA Feedback Enhancement

作　　者：吕闯 庞敏 刘普 LYU Chuang;PANG Min;LIU Pu(Guangdong Midea Air-Conditioning Equipment Co.,Ltd.,Foshan,Guangdong Province,528300 China)

机构地区：[1]广东美的制冷设备有限公司,广东佛山528300

出　　处：《科技资讯》2024年第23期48-53,共6页Science & Technology Information

摘　　要：为了解决传统空调温度控制方法在满足不同用户、不同场景需求时的局限性,提出了一种空调智能温度控制算法。此算法通过结合多元线性回归(Linear Regression)的趋势预测、贝叶斯网络(Bayesian Network)的室内外环境突变修正和决策树(Decision Tree)的用户体征修正,形成了一个综合考虑多种因素的温度调节模型,再通过三合一修正的加权平均(Weighted Average)方法,成为LBD-WA反馈提升模型。该模型不仅能够考虑实时的环境变化,还可以根据用户的温感和所处时区季节做出相应调整,从而更准确地为用户提供舒适的温度环境。为了验证该算法的效果,将此模型与其他机器学习方法进行了对比实验。结果显示:LBD-WA模型在预测准确性和多场景适应性方面表现优异,LBD-WA模型在准确率方面比传统空调方法提高了6%,这为未来智能家居的温控技术提供了一个新的方向。This article proposes an intelligent temperature and air conditioning control algorithm to address the limitations of traditional air conditioning temperature control methods in meeting the diverse needs of different us-ers and scenarios.The algorithm combines trend prediction using Multiple Linear Regression,environmental change correction using Bayesian network,and user characteristic correction using Decision Tree to form a tem-perature adjustment model that comprehensively considers multiple factors.By using the Weighted Average method with three in one correction,and becomes the LBD-WA feedback enhancement model.It not only considers real-time environmental changes but also adjusts according to the user′s temperature preferences and seasonal changes in their time zone to provide a more accurate and comfortable temperature environment.To validate the effectiveness of this algorithm,comparative experiments were conducted between this model and other machine learning meth-ods.The results show that the LBD-WA model performs well in terms of prediction accuracy and adaptability to multiple scenarios,with a 6%improvement in accuracy compared to traditional air conditioning methods.This pro-vides a new direction for temperature control technology in future smart homes.

关 键 词：多重反馈修正 多元线性回归 贝叶斯网络 决策树 异常调节反馈 时区季节反馈 体征变化反馈 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]