The invention discloses a power station boiler balanced combustion optimization method based on data driving, which includes a front-end sensor, a data model, a balanced combustion control center, flow velocity adjusting equipment, combustors and a flow velocity and flow measurement sensor, wherein the front-end sensor is used for detecting parameters reflecting the combustion state in a hearth, the data model analyzes and judges whether deviated combustion and insufficient combustion exist in the hearth or not, the balanced combustion control center receives output of the data model and air powder flow velocity and flow parameters measured by the flow velocity and flow measurement sensor, then analyzes and calculates an adjusting instruction, and sends the adjusting instruction to the flow speed adjusting equipment, and the flow speed adjusting equipment can change the resistance of a primary air pipeline and adjust the air powder flow speed and flow of the corresponding combustor, so that air powder combustion of all the combustors is optimized, sufficient combustion is achieved, and the combustion efficiency is improved. Primary air powder deviation can be adjusted on line under data driving, deviation combustion in the furnace is effectively solved, the combustion efficiency is improved, and coal consumption is reduced.