内蒙古西部地区民居被动式节能改造研究——以鄂尔多斯为例  

Passive Energy Saving Rentrofit of Residential Buildings in Western Inner Mongolia:Taking Ordos as an example

作　　者：王蒙 任玉成 张宝庆[2] 高小聪 徐昊 吉国华[1] 马强[3,4] WANG Meng;REN Yucheng;ZHANG Baoqing;GAO Xiaocong;XU Hao;JI Guohua;MA Qiang(School of Architecture and Urban Planning,Nanjing University,Nanjing 210000,China;School of Water Conservancy and Architecture Engineering,Shihezi University,Shihezi 832003,Xinjiang Uygur Autonomous Region,China;Department of Architecture and Engineering,Ordos Vocational College,Ordos 017000,Inner Mongolia Autonomous Region,China;Engineering Thermophysics Research Center,North China Electric Power University,Beijing 102206,China;School of Civil Engineering,Xinjiang University,Urumqi 830000,China)

机构地区：[1]南京大学建筑与城市规划学院,南京210000 [2]石河子大学水利建筑工程学院,新疆维吾尔自治区石河子832003 [3]鄂尔多斯职业学院建筑工程系,内蒙古自治区鄂尔多斯017000 [4]华北电力大学工程热物理研究中心,北京102206 [5]新疆大学建筑工程学院,乌鲁木齐830000

出　　处：《建筑节能（中英文）》2024年第6期87-93,共7页Building Energy Efficiency

基　　金：鄂尔多斯市应用技术研究与开发基金资助项目(2022YY020);内蒙古自治区教育科学基金资助项目(NJZY21167);江苏建筑节能与建造技术协同创新中心资助项目(SJXTGJ2104)。

摘　　要：目前,严寒地区的农村民居存在着能耗高、舒适度低的问题,相关研究较多集中在东北地区和内蒙古东部,而有关内蒙古西部的草原民居研究相对较少。针对这一现状,提出了一种针对内蒙古西部气候特征和民居问题的优化改造框架,该框架以能源使用强度(EUI)、热舒适度不满意者的百分数(PPD)、改造后全生命周期的成本(LCC)为性能目标,利用深度神经网络(DNN)和遗传算法(NSGA-II)对民居的围护结构进行优化。选取鄂尔多斯地区的一户草原民居进行案例研究,利用Grasshopper平台进行性能模拟并获取数据集,使用Python语言训练深度神经网络模型,最终得到R^(2)评价指标较为理想的预测模型,之后利用遗传算法对该模型进行优化得到多个帕累托前沿解,使用Kmeans聚类算法对帕累托前沿解进行分析选出自己理想的方案。最后将选出的方案和改造前的方案进行对比,结果表明:LCC、EUI、PPD分别降低35%、33%和17.75%。Rural dwellings in cold regions features for high energy consumption and low comfort.The relevant researches are mainly concentrated in northeast China and eastern Inner Mongolia,while the researches on grassland dwellings in western Inner Mongolia are relatively few.In view of this situation,this paper proposes an optimization and retrofit framework for the climate characteristics and residential problems in western Inner Mongolia.The framework takes Energy Use Intensity(EUI),Predicted Percentage of Dissatisfied(PPD),and Life Cycle Cost(LCC) after retrofitting as its performance objectives.Deep Neural Network(DNN) and Non-dominated Sorting Genetic Algorithm(NSGA-II) were used to optimize the residential building envelope.This paper selects a grassland residence in Ordos for case study,uses Grasshopper platform for performance simulation and data set acquisition,uses Python language to train the deep neural network model,and finally obtains a prediction model with relatively ideal R^(2) evaluation index.Then,genetic algorithm is used to optimize the model to obtain multiple Pareto frontier solutions.Kmeans clustering algorithm is used to analyze the Pareto frontier solution and select the ideal solution.The results show that LCC,EUI and PPD are reduced by 35%,33% and 17.75% respectively.

关 键 词：改造框架 性能优化 民居 

分 类 号：TU17[建筑科学—建筑理论] TU205