Research On Green Design Optimization Of Traditional Residential Houses In Northern Guilin Based On Parameterization And Machine Learning

Traditional Chinese construction activities carry with them an indigenous green ideology,and residents in different parts of the country have created traditional buildings that are energyefficient and liveable according to local conditions,with Jiangtou and Changgangling villages in northern Guilin being typical examples.This green building strategy is relevant to the current development of green buildings.Combining it with the currently developed parametric digital design concept and the idea of passive design for green buildings will further deepen the green design and optimization methods for buildings,profoundly affecting the energy loss of buildings and the superior experience of users.Therefore,it is necessary to investigate the wisdom of traditional architecture in the northern region of Guilin,explore the laws of building and climate adaptation,and combine them with digital construction technology to carry out indepth research in parametric digital design.This study aims to investigate the climatic adaptation of traditional dwellings in northern Guilin using a combination of parametric platforms and performance simulation tools.A combination of field and literature research was used to explore the construction patterns of traditional dwellings in northern Guilin and to extract design parameters and their value ranges.Based on Grasshopper,a parametric design platform,and the construction rules and parameters of traditional dwellings in northern Guilin,a parametric generation module was successfully built for traditional dwellings in northern Guilin.The multi-objective optimization module was built based on the Octopus plug-in.These three modules together form a parametric research platform for climate adaptation of traditional houses in northern Guilin.Based on this platform,a multi-objective optimization study is carried out on the performance of the light and thermal environments of traditional dwellings in northern Guilin to obtain the optimal set of values for key design parameters,thus achieving the optimization of the design objectives;finally,a design database for traditional dwellings in northern Guilin is built based on the design parameters and their performance simulation index values for climate adaptation in the solution set.The database was used to establish the quantitative relationship between the design parameters of traditional buildings and climate characteristics adaptation and green building performance indexes using machine learning(KNN,DT,RF,GBDT,XGBoost,and other algorithms).