Analysis and application of human thermal comfort modeling in hot summer and cold winter areas

When the indoor thermal environment changes, the human body adapts to these changes through a series of physiological adjustments, so physiological parameters are also used as indicators to evaluate thermal comfort. However, research on automated judgment of thermal comfort is still not perfect. As thermal sensation is an important expression of thermal comfort, this study selected an environment of 19.6 °C -32.5 °C for the experiments. The results indicate that there is the strongest correlation between hand temperature and thermal sensation. When there is only one feature, selecting the hand as the feature input, the accuracy of LR (Logistic Regression) and SVM (Support Vector Machine) models reaches 0.72. When there are many features, selecting eyes, cheeks, hands or eyes, nose, cheeks, hands as feature inputs, the accuracy of using the random forest model reaches 0.8. Finally, the use of images effectively identifies body parts and obtains temperature, achieving automated thermal comfort prediction. To provide some theoretical support for thermal comfort research through the study of human temperature and thermal comfort.