摘要
综述了建筑方案能耗快速预测的4种方法:工程简化算法、以多元回归方法为代表的统计学方法、以人工神经网络模型为代表的人工智能方法和并行计算方法,并从预测精度、预测速度、方法适用情况、方法局限性等方面进行了对比,为针对具体问题的预测方法选择给出了建议。
Reviews four methods of rapid prediction of building energy consumption including the engineering simplification algorithm, the statistical method represented by multiple regression method, the artificial intelligence method represented by artificial neural network model and the parallel computing method. And performs the comparison from the aspects of prediction accuracy and speed, method application and limitation, and so on. Gives the suggestions for the selection of prediction method for specific problems.
作者
李紫微
林波荣
陈洪钟
Li Ziwei;Lin Borong;Chen Hongzhong(Tsinghua University, Beijing, Chin)
出处
《暖通空调》
2018年第5期1-8,共8页
Heating Ventilating & Air Conditioning
基金
国家自然科学基金资助项目"中英低碳建筑整体性能研究"(编号:NSFC51561135001)
"十三五"国家重点研发计划课题"建筑全寿命期绿色性能模拟分析技术与新型设计工具"(编号:2016YFC0700209)
关键词
建筑能耗预测
工程简化算法
多元回归方法
人工神经网络模型
并行计算方法
building energy consumption prediction, engineering simplification algorithm, multiple regression method, artificial neural network model, parallel computing method
