To address the systematic limitations of traditional optimization methods for coal-fired boilers and the inability to accurately perceive in-furnace combustion states, this paper proposes a comprehensive boiler optimization system integrating intelligent sensing and control. First, an infrared temperature measurement array is deployed, combined with a gradient positioning algorithm, to achieve online reconstruction and visualization of the three-dimensional temperature field within the furnace. Second, a dynamic model of the boiler combustion process is established based on a continuous-time Bayesian network. Finally, a multi-objective particle swarm optimization algorithm with dynamically adjusted inertia weights is employed for online optimization, thereby constructing a real-time closed-loop adaptive intelligent combustion control system. Engineering application results demonstrate that the proposed system can effectively perceive the combustion state and accurately identify and provide early warnings for abnormal conditions, such as slagging and uneven combustion. After the system was put into operation, the boiler efficiency increased by no less than 0.3%, and NOx emissions were reduced by no less than 12%. In conclusion, this system provides robust technical support for resolving operational optimization challenges in utility boilers and achieving the synergistic development of safety, economic efficiency, and environmental protection.