Abstract: For the dynamic characteristics of a superheated steam temperature of a highly non-linear thermal power production process with a large delay and inertia, a superheated steam temperature control system is proposed with an adaptive Smith predictor. This system is optimized to improve the dynamic model structure of the closed-loop system. It has a simple structure, is convenient when adjusting parameters, and is easy to realize the configuration of the distributed control system. Simulation results show that both the tracking performance of set-point values and the anti-interference ability of the system are significantly better than those of systems such as the classical cascade control system for the steam temperature and the adaptive fuzzy control system for the boiler superheated steam temperature. This improvement is based on the optimization of controller parameters and the composited fuzzy control system with an improved Smith predictor.