Machine learning-based multi-objective optimisation of an aerogel glazing system using NSGA-II—study of modelling and application in the subtropical climate Hong Kong

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

7 Scopus Citations
View graph of relations

Author(s)

Detail(s)

Original languageEnglish
Article number119964
Journal / PublicationJournal of Cleaner Production
Volume253
Online published6 Jan 2020
Publication statusPublished - 20 Apr 2020

Abstract

Application of super-insulating materials in building glazing system shows promising prospects for low-energy buildings. In this research, the heat transfer, solar radiation transmission and indoor illuminance of an aerogel glazing system were characterized through an experimentally validated numerical model. Contribution ratios of multi-variables to multi-objectives were thereafter quantified, following the Taguchi standard orthogonal array. In respect to the application of aerogel glazing system in subtropical climates, an energy-related contradiction between indoor illuminance from solar and indoor heat gain, has been presented, discussed, together with effective solutions. In order to minimise the total heat gain and maximise the indoor illuminance transmitted through the aerogel glazing system, a generic multi-objective optimisation methodology, with high computational efficiency and accuracy, has been developed, to identify the optimal design. The results indicate that through the Pareto front from the multi-objective optimisation results, a significant reduction of total heat gain and an obvious increase of the indoor illuminance can be noticed. Compared to the optimal case in the standard orthogonal array, with the application of the proposed multi-objective optimisation methodology, the annual total heat gain could be reduced from 489305.5 to 333396.4 Wh by 31.9% and the annual indoor illuminance could be increased from 56786.6 to 172973.5 lux by 67.2%. The year-round performance indicates that, compared to the bi-objective optimisation (annual transmitted heat gain and annual indoor illuminance) with the annual total heat gain at 333.4 kWh/m2 and annual indoor illuminance at 162.3 klux, the bi-objective optimisation (annual total heat gain and annual indoor illuminance) shows a lower annual total heat gain at 322.4 kWh/m2 by 3.4%) and a higher annual indoor illuminance at 173 klux by 6.6%. This study proposes an overall framework and technical guidance of a new multi-objective optimisation methodology, which can automatically learn mechanisms of heat transfer and solar radiation transmittance through nanoporous aerogel granules, and identify the optimal multi-variables setting for the robust system design and operation.

Research Area(s)

  • Aerogel glazing system, Indoor heat gain, Indoor illuminance, Machine learning, Multi-criteria decision making, Multi-objective optimisation

Citation Format(s)

Machine learning-based multi-objective optimisation of an aerogel glazing system using NSGA-II—study of modelling and application in the subtropical climate Hong Kong. / Zhou, Yuekuan; Zheng, Siqian.

In: Journal of Cleaner Production, Vol. 253, 119964, 20.04.2020.

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal