The increasing use of transparent envelope in building enclosure will lead to high energy consumption,thermal discomfort and other problems.In order to solve these problems,shading is used more and more as an effective means to reduce energy consumption of buildings and improve indoor environment.In order to explore the influence degree of louver external shading parameters on shading performance,machine learning algorithm was used to predict shading performance.Improved sensitivity analysis method based on machine learning was used to discuss the local and global sensitivity of two types of parameters(building and shading)affecting shading performance,and to determine the most influential parameters.The results show that XGBoost has the highest accuracy in predicting thermal environment and energy consumption indexes,while random forest algorithm has the best effect in predicting optical environment indexes.Meanwhile,it is found that shading parameters are the most important factors affecting indoor thermal environment and building energy consumption,and the overall weight is above 0.5.Building parameters significantly affect indoor lighting,with a weight of about 0.9.
NETL
NSTL
万方数据
