一种基于数据驱动的空调负荷预测方法

摘要：

空调负荷预测是空调负荷潜力分析和电网空调负荷调控的基础，为了精确地对空调负荷进行预测，文中提出了一种考虑到外界影响因素以及集成优化的空调负荷预测方法。首先，拟定好实验运行方案并采集影响因素数据。其次，使用近邻成分分析（NCA）方法进行特征选择，剔除重要度小的特征。然后使用白鲨优化算法（white shark optimizer，WSO）对支持向量回归（supportvector regression，SVR）的正则化参数和核函数的宽度参数进行优化，最后，结合自适应提升算法（adaptive boosting，Adaboost）构建Adaboost-WSO-SVR主模型，检验其精度并与其他方法进行比较。结果表明，提出的 Adaboost-WSO-SVR主模型相比于集成优化后的BP，ELM 模型精度更高。可知提出的方法在负荷预测方面效果更好，为空调节能优化控制策略提供依据。

Air conditioning load forecasting is the basis for analyzing the potential of air conditioning load and regulating the air conditioning load of the power grid. In order to accurately predict the air conditioning lad, this paper proposes an air conditioning load forecasting method taking into account of external influencing factors and integrates optimization, Firstly, develop an experimental operation plan and collect data on influencing factors, Secondly, the nearest neighbor component analysis(NCA) method is used for feature selection to remove features with low importance, Then the white shark optimizer(WS0) algorithm for support vector regression(SVR)are used.The regularization parameter of SVR and the width parameter of the kernel function are optimized, and finally, the adaptive boosting algorithm is combined, Construct the Adaboost WSO-SVR main model, test its accuracy, and compare it with other methods. The results indicate that the accuracy of the Adaboost WSO-SVR main model proposed in this article is higher than the integrated optimized BP, ELM models. It is known that the proposed method has better performance in load forecasting, providing a basis for optimizing control strategies for air conditioning energy

conservation.

作者：

周孟然，周光耀，胡锋，朱梓伟，张奇奇，王玲，孔伟乐，吴长臻，崔恩汉

Zhou Mengran, Zhou Guangyao, Hu Feng, Zhu Ziwei, Zhang Qiqi, Wang Ling, Kong Weile, Wu Changzhen, Cui Enhan

机构地区：

安徽理工大学电气与信息工程学院

引用本文：

周孟然，周光耀，胡锋等。一种基于数据驱动的空调负荷预测方法 [J] . 学报(自然科学版) ,2025, 53(3) :128-134. (Zhou Mengran,Zhou Guangyao, Hu Feng,et al.A data-driven method for air conditioning load forecasting [J] .Journal of Henan Normal University(Natural Science Edition) , 2025, 53(3) : 128-134. DOI: 10.16366/j. cnki.1000-2367. 2024. 06. 23. 0001. )

基金：

国家自然科学基金；安徽省自然科学基金能源互联网联合基金重点项目；国网安徽省电力有限公司阜阳供电公司科技项目

关键词：

空调负荷；负荷预测；特征选择；白鲨优化算法；自适应提升算法；支持向量回归

air conditioning load: load forecasting; feature selection; white shark optimization algorithm ; adaptive boosting algorithm; support vector regression

分类号：

TU831

一种基于数据驱动的空调负荷预测方法.pdf