Thermal Comfort Modeling of Office Buildings Based on Improved Random Forest Algorithm

HVAC systems usually regulate the thermal comfort of the building environment according to standards. However, residents living in different climatic conditions have different thermal preferences, unified standards for HVAC regulation may cause cold and thermal discomfort of indoor personnel and lead to excessive energy consumption. In order to analyze the thermal comfort of indoor personnel more accurately, this paper proposed an indoor thermal comfort model of office buildings based on an improved random forest algorithm. First, by using the data from a building environment smart monitoring system in an office environment, the thermal comfort is evaluated by experimenters in a variety of thermal environments. Then, based on the experimental results, K-means algorithm is used to select the decision tree with low similarity, to form a new random forest, which has higher prediction accuracy for the test set. Finally, our proposed model is embedded in the smart monitoring system to verify its advantaged and effectiveness. The experimental results show that the model is suitable for describing the thermal comfort of personnel in office space, which has higher precision and good generalization ability, and can be further deployed to HVAC system.