On-line diagnosis of boiler combustion in power plants is of great significance to the construction of smart power station. Realizing automatic close-loop control for safety

stable and excellent combustion through smart diagnosis is one of the key issues that need to be solved for smart operation of power units. In this paper

the boiler combustion status is diagnosed online by using the fuzzy C mean clustering algorithm based on flame intensity signal. Via analyzing the history data of combustion states

four categories of combustion state are summarized and the cluster center of each category is obtained. The combustion state of a flame is determined by calculating the membership between the diagnosing sample and the cluster center of each category. Moreover

this method is practically applied on a supercritical 620 MW unit to analyze the diagnosis results under four typical conditions

namely the steady output of the burner

varying output of the burner

unstable output of the burner and the condition from ignition to steady output of the burner. The results show that

the method proposed in this work is probable to accurately estimate the combustion state of all conditions

and the diagnosing conclusion reflects objectively the actual combustion process. Based on the online diagnosis

it is able to provide real-time guidance for combustion optimization.