期刊
JOURNAL OF COMPUTATIONAL DESIGN AND ENGINEERING
卷 9, 期 5, 页码 2103-2133出版社
OXFORD UNIV PRESS
DOI: 10.1093/jcde/qwac100
关键词
metaheuristic optimization; hybrid algorithm; architectural design optimization; building performance simulation
资金
- National Research Foundation of Korea (NRF) - Korea government (MSIT) [NRF-2021R1C1C1003403, NRF-2019R1A2C100913012]
- Ajou University Research Grant [S-2021-G0001-00016]
This paper proposes a novel metaheuristic optimization technique DECS-TAPS for rapid and optimal digital prototyping of architectural forms. The findings show that DECS-TAPS outperforms other algorithms in multi-objective optimization and computational effectiveness, while also reducing parameter dependence and increasing robustness.
One of the critical limitations in architectural design optimization (ADO) is slow convergence due to high-dimensional and multiscale variables. For the rapid and optimal digital prototyping of architectural forms, this paper proposes a novel metaheuristic optimization technique that hybridizes standard low-level algorithms: the differential evolutionary cuckoo-search-integrated tabu-adaptive pattern search (DECS-TAPS). We compared DECS-TAPS to 10 major standard algorithms and 31 hybrids through 14 benchmark tests and investigated multi-objective ADO problems to prove the computational effectiveness of multiple algorithm hybridization. Our findings show that DECS-TAPS is vastly efficient and superior to the covariance matrix adaptation evolution strategy algorithm in multifunnel and weak structural functions. The global sensitivity analysis demonstrated that integrating multiple algorithms is likely conducive to lowering parameter dependence and increasing robustness. For the practical application of DECS-TAPS in building simulation and design automation, Zebroid-a Rhino Grasshopper (GH) add-on-was developed using IronPython and the GH visual scripting language.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据