分类
多目标优化
遗传算法
建筑信息建模
参数统计
工程类
空调
数学优化
包络线(雷达)
可靠性(半导体)
算法
计算机科学
模拟
机器学习
数学
机械工程
电信
雷达
统计
功率(物理)
物理
量子力学
调度(生产过程)
作者
Yang Liu,Tiejun Li,Wenbo Xu,Qiang Wang,Hao Huang,Bao‐Jie He
标识
DOI:10.1016/j.enbuild.2023.113665
摘要
Green buildings (GB) have been widely promoted in various nations. However, the post occupancy evaluation suggests many GB cannot well fulfill the expected targets. To overcome the mismatch among GB’s multi-objectives, this paper develops an efficient and intelligent hybrid method using BIM-DesignBuilder (BIM-DB), Grey wolf optimization (GWO), random forest (RF) and non-dominated sorting genetic algorithm II (NSGA-II) to achieve the optimization of design parameters. The BIM model and DB simulation tool were used to obtain data samples of envelope and air conditioning system design parameters, and their life cycle carbon emission (LCCE), economic cost (EC) and predicted mean vote (PMV). The RF model was used to achieve high precision prediction. The GWO was used in the hyper-parameter optimization. The NSGA-II algorithm was applied to multi-objective optimization to obtain optimal design parameters. A building case shows: (1) The RF model had an excellent prediction performance for LCCE, EC and PMV. (2) BIM-DB can be used to obtain low error and high reliability building simulation data sets. (3) The RF-NSGA-II intelligent algorithm can reduce the LCCE of the building in the entire cycle by 16.6%, reduce the EC per square meter by 2.0%, and greatly improve the thermal comfort by 18.3%, representing good application value. This research provides a way of thinking for the multiobjective optimization of green buildings from the perspective of data mining and guidance for the parameter selection of the envelopes and air conditioning systems of new and existing buildings to more scientifically and effectively design green buildings.
科研通智能强力驱动
Strongly Powered by AbleSci AI