Research on Intelligent Control of Air Conditioning System Based on Raspberry PI and Deep Reinforcement Learning

This study addresses the gap in research on human thermal comfort in intelligent air conditioning systems by proposing an innovative optimization control framework integrating Internet of Things (IoT) and artificial intelligence (AI) technologies. Utilizing a Raspberry Pi-based data acquisition system to gather comprehensive environmental and physiological parameters, high-precision predictive models for both thermal comfort and energy consumption were developed through rigorous algorithm comparison. The ensemble learning approach augmented with infrared thermal imaging demonstrated superior performance in thermal comfort prediction, while the gated recurrent unit (GRU) model exhibited distinct advantages in energy consumption forecasting. To achieve optimal system performance, a deep reinforcement learning-based (DQN) intelligent control strategy was developed, successfully coordinating thermal comfort and energy efficiency. Experimental results show the proposed method maintains user comfort while achieving a 9% reduction in energy consumption, presenting a novel technical solution for advanced air conditioning system control.