Abstract: To address issues of traditional winter heating methods in facility agriculture, such as high energy consumption, insufficient temperature control accuracy, and severe pollution, a low-energy automation greenhouse heating control system based on methanol combustion has been designed and developed．This system has integrated a multi-sensor data acquisition network capable of monitoring indoor and outdoor temperature, humidity, CO₂ concentration, and light intensity．Based on this, a dynamic greenhouse temperature prediction model utilizing machine learning algorithms has been constructed．Furthermore, a fuzzy PID control strategy has been introduced to enable predictive control of methanol combustion heating system, effectively overcoming issues of temperature lag and high energy consumption associated with traditional heating methods．Additionally, this system has innovatively implemented a closed-loop resource utilization system for CO₂ generated from methanol combustion, and has established comprehensive remote monitoring and multi-layered safety protection system．Practical application results have demonstrated that this system can provide crops with a stable and suitable growing environment．While achieving energy savings, it significantly improves crop yield and quality, offering a comprehensive solution for intelligent upgrading and green, low-carbon development of facility agriculture．