A three-stage optimization method for the classroom envelope in primary and secondary schools in China. (15th July 2022)

• Record Type: Journal Article
• Authors: Xu, Yizhe; Yan, Chengchu; Pan, Yan; Zhao, Ke; Li, Mei; Zhu, Faxing; Jiang, Yanlong
• Is Part Of: Journal of building engineering. Volume 52(2022)
• Publication Date: 2022-07-15
• Subjects: Envelope optimization -- Hybrid algorithm -- MFNN -- School classroom -- Algorithm comparison; Building -- Periodicals; 690.05
• DOI: 10.1016/j.jobe.2022.104487
• Languages: English
• ISSNs: 2352-7102

Abstract

Abstract: The design optimization of primary and secondary school classrooms should consider the effects of energy, thermal comfort, and daylighting. This paper proposes a three-stage optimization design method of classroom envelope based on multi-objective optimization to realize the efficient and accurate optimization of primary and secondary school classrooms. The method is divided into three modules: building modeling and analysis, meta-model training and hyperparameter optimization, and multi-objective optimization algorithm comparison. The application effect of electrochromic technology in typical primary and secondary school classrooms is discussed through sensitivity analysis, and an improved algorithm SEGA-Q based on GA is proposed. In addition, the performance of three typical multi-objective optimization algorithms in three objective optimizations (i.e., energy, thermal comfort, and daylighting) is compared. Finally, two typical school classrooms are taken as cases to verify the effectiveness of the proposed optimization method. The results show that the application significance of the electrochromic window indicates skylights > south side window > north side window. The SEGA-Q algorithm has more accuracy, stability, and convergence speed advantages in meta-model training and hyperparametric optimization, but its calculation time increases significantly. Among the three multi-objective optimization algorithms, the NSGA-II algorithm has the most significant advantages in most aspects. In addition, compared with the benchmark schemes, the optimized common classroom can reduce the TES by 692.5 kWh at most and increase the PT by 41.9% and UDI NO by 57.4%. The optimized classroom with skylight can reduce TES by 1261.9 kWh at most and increase PT by 26% and UDI NO by 78.5%. Highlights: A three-stage optimization design method of the classroom envelope is presented. An improved algorithm SEGA-Q based on GA is proposed. The effects of electrochromic window in different positions are discussed. The effects of GA and SEGA-Q on MFNN model optimization are compared. The performances of three multi-objective optimization algorithms are compared.