Under the premise of safe operation, to improve the economy and environmental protection of thermal power units, a combustion optimization framework is proposed, based on the historical operation data, combustion adjustment data and real-time running data. In this framework, steady-state detection, on-line calculation of boiler efficiency, data mining, nonlinear modeling, intelligent optimization and other techniques are adopted, and basic control references and real-time control increments are obtained. After reliable communication and undisturbed configuration with DCS, these results can participate in the real-time control optimization of boiler combustion, according to the running state of the thermal power unit. It has been applied in coal-fired power plant, realizing analysis of combustion states, automatic optimization of setpoints and improvement of boiler efficiency in a wide range of load. Based on the framework, the technical elements can be expanded and optimized, and it can be applied to various types of boilers and control systems.