An integrated framework for residential layout designs: Combining parametric modeling, neural networks, and multi-objective optimization for outdoor activity space optimization

Zhuoran Hu,Lingqing Zhang,Qiong Shen,Xiaowei Chen, Wenjing Wang, Kunpeng Li

ALEXANDRIA ENGINEERING JOURNAL(2023)

引用 0|浏览0
暂无评分
摘要
The quality of life of residents and community vitality is significantly enhanced by outdoor activity spaces within residential areas. Existing residential layout designs primarily focus on a restricted set of predefined layouts and most outdoor evaluation indices are limited to thermal comfort, thereby leading to an incomplete evaluation system. Furthermore, there is a scarcity of studies offering a comprehensive design process, from parameterized generation to performance evaluation. As such, there is a pressing need for an all-encompassing design framework that investigates a broad array of layout possibilities while providing multidimensional evaluations based on diverse metrics, thereby augmenting the accuracy and intelligence of residential layout designs.This research introduces a building layout generation framework that integrates parametric modeling (PD), neural network modeling (ANN), and multi-objective optimization (MOO) methods. The framework utilizes parametric design methods to construct a geometric model that adheres to building codes. Following this, it builds separate neural network models for three key metrics: sky view factor (SVF), sunshine duration, and noise, this approach is aimed at accelerating the evaluation process. This alternative model serves as the objective function for multi-objective optimization, identifying the optimal building layout that balances multiple metrics. Additionally, the framework utilizes the ideal point method to optimize the final Pareto solution set, thereby informing designers' and planners' program selections.Case studies of this framework indicate that the residential layouts generated can adhere to building codes while effectively balancing the three crucial metrics. Relative to the conventional modeling-evaluation process, this framework augments efficiency substantially. The design framework presented herein will offer comprehensive decision support for policy makers to optimize residential layout designs, further fostering the harmonious coexistence of people and cities, while advancing intelligent building design and sustainable urban development.
更多
查看译文
关键词
Multi-objective optimization,Machine learning,Building layout,Outdoor environment
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要