Heat comfort in passenger automobile interiors is a major issue affecting human health, safety and driving ability, especially in hot climatic areas. Heat in the cabin might cause heat stress, fatigue, discomfort, and impaired thinking and thus reduce the effectiveness of driving and risk of accidents. Most of the current studies are based on laboratory experiments or computational fluid dynamics (CFD) simulation either without considering the actual conditions of driving or the surface heat transfer of the vehicle body, or without direct human physiological reactions. This article presents a sensor-based machine learning system in the real world to forecast the human thermal comfort and heat stress in the cabins of passenger cars. Under real driving conditions, the environmental parameters, along with the individual vehicle body temperatures and the physiological indicators of human beings, such as the heart rate and skin temperature. Random Forest, XGBoost and Long Short-Term Memory (LSTM) network machine learning models are developed and tested. The findings prove that vehicle body heat inclusion and physiological parameters inclusion significantly increase the accuracy of predictions. The suggested framework can, therefore, provide a workable intervention towards intelligent cabin climate control and driver safety enhancement in hot weather.
Thakur et al. (Thu,) studied this question.