Abstract: With the continuous upgrading of air conditioning usage demand, people's attention to the comfort and energy-saving performance of room air conditioners and other equipment is also increasing. Six key indicators for indoor thermal comfort evaluation are integrated: average thermal sensation index, thermal sensation voting, effective temperature, standard effective temperature, air distribution characteristic indicators, and thermal comfort voting model. These indicators are divided into two categories: subjective evaluation models for thermal comfort and objective evaluation models for thermal comfort. The objective evaluation model can be calculated through specific formulas, while the subjective evaluation model relies on people's voting results. However, neither of these models can currently be directly applied to the control of air conditioning systems. To address this issue, two approaches are summarized to solving it. The first approach is to simplify the objective evaluation model for thermal comfort, so that the air conditioning system can be effectively controlled based on the simplified model. The second approach is to use a subjective and objective evaluation model for thermal comfort to establish a machine learning model. This model will be driven by the data collected by sensors and the image data obtained by thermal imaging technology to achieve accurate control of the air conditioning system. Through these two methods, the comfort and energy-saving performance of the air conditioning system can be significantly improved, meeting people's pursuit of high-quality life.

Key words: Thermal comfort, Air conditioning control, Model simplification, Data driven