DINOv2-Assisted Selection and Life Cycle Cost Control of Green Building Textiles

Hangtian Liu

Article
2026 / Volume 9 / Pages 828-848
Received 30 July 2025; Accepted 24 September 2025; Published 31 March 2026
https://doi.org/10.31881/TLR.2026.828

Abstract
The lack of selection accuracy and poor life cycle cost control of functional textile materials in green buildings re-stricts their performance in energy savings, safety, environmental protection, and other performance indicators. This paper implements a method for intelligent selection and life cycle cost control of building textile materials that integrates the DINOv2 model. First, a local texture feature extraction module is introduced into DINOv2 to enhance its modeling capabilities for high-frequency textures and complex tissue structures, enabling image-level semantic recognition of the sustainable performance of textile materials. Subsequently, a multi-factor regression model is constructed to quantify the comprehensive cost of textile materials by combining functional parameters such as material heat and moisture regulation, flame retardancy, durability, and economic data. Finally, an adap-tive mutation rate mechanism is introduced into the genetic algorithm (GA) to perform multi-objective optimiza-tion on material combinations that meet the performance constraints of green buildings. Experimental results show that the proposed method can complete optimization in 63.1 seconds for the selection of 80 types of textile materials, achieving the lowest life cycle cost of 295,000 CNY and a performance score of 94, providing an effec-tive solution for the selection of textile materials and full life cycle cost control in green buildings.

Keywords
green building, textile material selection, full life cycle cost control, DINOv2 model