The small biomass boiler heating system (SBBHS) offers a cost-effective, convenient, safe, and environmentally friendly heating solution for small-scale users, providing notable social and economic advantages. Temperature prediction and control methods can enable SBBHS to operate more intelligently and autonomously, further minimizing heating expenses. This study focuses on a small biomass boiler heating system in Xinyang, Shandong, utilizing data-driven methods to analyze SBBHS performance in supply water temperature prediction and optimization. To achieve precise temperature predictions, an enhanced artificial neural network model is developed, trained, and validated, with the Levenberg-Marquardt optimization algorithm applied to adjust network weights and thresholds. Additionally, a feedback neural network is employed for short-term, 24-hour temperature predictions of the SBBHS. Experimental results demonstrate that this temperature prediction and control strategy ensures long-term indoor temperature stability and comfort while reducing heating costs. This research contributes to the intelligent upgrading and transformation of small biomass boiler control systems, enabling on-demand heating and reducing carbon emissions.