Furnace temperature control has a direct impact on steel qualities and energy consumptions in a steel rolling process. Due to dramatic changes of combustion, this multi-variable process is time-varying, and also has inherent nonlinearities, couplings among the variables, large inertias and time delays. Therefore, manual operations are still being widely used in the furnace temperature control. In this paper, a two-layer control structure is proposed for furnace temperature control of a steel rolling process. This structure consists of an upper layer to set the desired temperature (set point) to be tracked by system output and a lower layer to stabilize the furnace temperature system. Here, will only introduce temperature set point.. First, according to Scheduling, Usage the second performance optimization methods to optimize the furnace operation indicators to determine the ideal billet heating process and to determine the furnace temperature set point; And then, the ideal billet heating curve for a given value, billet temperature prediction model estimates the value for feedback of PID feedback control, readjust set point value of the furnace temperature; and then, compensates the accuracy of prediction model using Billet temperature deviation between it is in rolling line and it is temperature baked. At the same time, the delay has been resolved. Finally, the real-time tracking module has been established, module mastering the location and temperature of billets in the furnace, corresponding the set point of the furnace temperature to be modify in real time to ensure optimization of performance indicators; The proposed hybrid intelligent control method has been successfully applied to some steel plant in Gansu Province, China.