计算机工程 ›› 2019, Vol. 45 ›› Issue (6): 280-289,296.doi: 10.19678/j.issn.1000-3428.0051781

• 开发研究与工程应用 • 上一篇    下一篇

基于遗传优化决策树的建筑能耗短期预测模型

丁飞鸿1a,1b,2,刘鹏1a,1b,2,卢暾1a,1b,2,顾宁1a,1b,2,丁向华1a,1b,2,杨宝明3,戴文祺3,邹超君3   

  1. 1.复旦大学 a.计算机科学技术学院; b.上海数据科学重点实验室,上海 201203;2.上海智能电子与系统研究院,上海 201203; 3.上海鲁班软件股份有限公司,上海 200433
  • 收稿日期:2018-06-11 出版日期:2019-06-15 发布日期:2019-06-15
  • 作者简介:丁飞鸿(1995—),男,硕士研究生,主研方向为协同计算、数据分析;刘鹏,博士;卢暾(通信作者),副教授、博士;顾宁,教授、博士、博士生导师;丁向华,副教授、博士;杨宝明,高级工程师、博士;戴文祺、邹超君,学士。
  • 基金项目:
    上海市科委项目“信息化技术与调适技术的集成研究与示范”(16DZ1202402)。

Short-term prediction model of building energy consumption based on genetically optimized decision tree

DING Feihong1a,1b,2,LIU Peng1a,1b,2,LU Tun1a,1b,2,GU Ning1a,1b,2,DING Xianghua1a,1b,2,YANG Baoming3,DAI Wenqi3,ZOU Chaojun3   

  1. 1a.School of Computer Science; 1b.Shanghai Key Laboratory of Data Science,Fudan University,Shanghai 201203,China;2.Shanghai Institute of Intelligent Electronics and System Research,Shanghai 201203,China; 3.Shanghai Luban Software Co.,Ltd.,Shanghai 200433,China
  • Received:2018-06-11 Online:2019-06-15 Published:2019-06-15

摘要: 使用线性回归模型预测公共建筑能耗数据时,存在不确定性影响因素和精度偏差问题。为此,建立一种遗传优化决策树模型。采用遗传算法优化梯度提升决策树的子树生成过程,以模型预测的R-Square值作为迭代的评估标准,从而达到能耗预测的目的。实验结果表明,与传统的回归预测模型相比,该模型预测精度较高。

关键词: 建筑能耗, 回归预测, 梯度提升决策树, 遗传算法, 遗传优化决策树

Abstract: When using linear regression models to predict public building energy consumption data,there are uncertainties and precision deviations.Therefore,a Genetically Optimized Decision Tree(GODT) model is established.The Genetic Algorithm(GA) is used to optimize the subtree generation process of the Gradient Boosting Decision Tree (GBDT).The R-Square value predicted by the model is used as the evaluation criterion of the iteration to achieve the purpose of energy consumption prediction.Experimental results show that the prediction accuracy of this model is higher than that of the traditional regression prediction model.

Key words: building energy consumption, regression prediction, Gradient Boosting Decision Tree(GBDT), Genetic Algorithm(GA), Genetically Optimized Decision Tree(GODT)

中图分类号: