The work aims to reduce the paper basis weight deviation by studying the large time delay control strategy in the paper basis weight control loop. On the basis of traditional Smith predictor and combined with fractional order PID algorithm, a two-degree-of-freedom Smith predictor composed of traditional PID controller and fractional order PID controller was designed to independently control set value tracking and load disturbance. Set point tracking adopted traditional Smith predictor. Fractional order PID controller with strong anti-interference and robustness was served for load disturbance. The control algorithm was applied in the paper basis weight and compared with the traditional control algorithm. With respect to traditional Smith control, the set value tracking and anti-interference of two-degree-of-freedom Smith predictor based on fractional order PID were more excellent. In the process of paper basis weight control, the paper basis weight deviation was reduced to ±1.6 g/m2. The two-degree-of-freedom Smith predictor based on fractional order PID can effectively control time delay. With its good robustness when the model is mismatched, the control accuracy of paper basis weight control loop can be effectively improved.