The value of back pressure of direct air-cooled units is mainly set by operating experience

which results in poor unit economy. Finding the best back pressure under different working conditions through reasonable methods is an effective measure to improve the economy of the unit. By using the gray correlation analysis method

this paper analyzes the main factors affecting the net power of the unit. Based on the operating data of a 600 MW direct air-cooled unit in a power plant

the random forest algorithm is used to establish the net power-back pressure model of direct air-cooled units at different load segments and different ambient temperatures. Through the model

the back pressure corresponding to the maximum net generating power of the unit under different working conditions is obtained

that is

the optimal back pressure. The study found that the optimal back pressure increases with the ambient temperature and load. The optimal back pressure is mostly smaller than the actual back pressure

and a proper reduction of the back pressure is beneficial to increase the net power of the unit. The net generating power increase opti mized by the optimal back pressure of the unit at low load is more significant.