基于遗传算法-阻容模型的建筑模型辨识及温度预测研究  

Research on building model identification and temperature prediction based on genetic algorithm and resistance capacitance model

作　　者：程亨达 刘佩晗 苟伟 喻方荣 陈焕新[1] 王江宇[1] Cheng Hengda;Liu Peihan;Gou Wei;Yu Fangrong;Chen Huanxin;Wang Jiangyu(Huazhong University of Science and Technology;China No.15 Metallurgical Construction Co.,Ltd.)

机构地区：[1]华中科技大学 [2]中国十五冶金建设集团有限公司

出　　处：《制冷与空调》2021年第9期15-20,共6页Refrigeration and Air-Conditioning

基　　金：国家自然科学基金(No.51876070,No.51576074)。

摘　　要：为实现精确预测建筑室内温度,得到最优控制策略,基于不同的简化假设,设计不同结构的建筑热阻容模型(RC模型),推导表征模型的微分方程组,在仿真结果的基础上采用遗传算法对模型进行辨识,得到一种兼顾模型预测精度和模型复杂度的RC模型结构。结果表明,在4种模型中,模型3对训练集温度的预测误差为0.346℃,对测试集温度的预测均方根误差为0.419℃,且模型结构较为简单,能够对建筑室内温度实现准确、稳定、相对快速的预测。In order to achieve accurate prediction of indoor temperature and obtain the optimal control strategy of building,several resistance-capacitance models(RC models)with different structures based on various simplifying assumptions are designed,and differential equations which characterize the models are deduced.Genetic algorithm is used to identify models based on simulation data to get parameters which make the model have best balance of prediction accuracy and structure simplicity.The result shows that among the four models,the RMSE of model 3 is 0.346℃on the train set,and 0.419℃on the test set.The structure of model 3 is relatively simple,and it can achieve accurate,stable forecast in relatively quick speed.

关 键 词：建筑节能 温度预测 热阻容模型 遗传算法 

分 类 号：TB6[一般工业技术—制冷工程]