Optimizing thermal comfort in highly automated vehicles: An AI-Based HVAC management approach with radiant panels for winter conditions

Abstract

This paper presents an AI-based model for optimizing heating, ventilation, and air conditioning (HVAC) settings to improve thermal comfort in electric vehicles under winter conditions and to estimate the associated power consumption. Unlike conventional HVAC systems that primarily rely on convective heating, the investigated concept combines convective airflow with nine radiant heating panels to enhance comfort and energy efficiency. Equivalent temperature (ET) was employed as an objective thermal comfort metric, and an XGBoost (Extreme Gradient Boosting) model was trained to predict ET for 16 body regions, achieving a high accuracy (coefficient of determination $R^2=0.96$). A Random Forest model was applied to relate fan speed and damper settings to mass flow. Validation experiments confirmed that the optimized HVAC settings maintained thermal comfort, with at least 50% of local body regions and 100% of upper and lower body averages within the neutral comfort zone. The approach demonstrated potential power savings of up to 240 W compared to convection-dominant strategies. These findings highlight the potential of combining AI with hybrid HVAC concepts to improve passenger comfort and reduce energy consumption in future automated electric vehicles.