In the response process of rapid peak regulation and frequency regulation of coal-fired units, quickly controlling the steam temperature within a reasonable range is one of the difficulties in the field of automatic control. Due to the large inertia and large delay characteristics of steam temperature, the conventional cascade control often fails to achieve better control results. Meanwhile, advanced algorithms such as model prediction and neural networks are affected by non-linearity, measurement uncertainty and time-varying characteristics of temperature object, which limits their application in actual control. This paper proposed a hybrid control model based on cascade and local Smith prediction. The accurate local Smith prediction model was obtained through parameter identification method, and the steam temperature control strategy based on the hybrid structure was constructed and applied in the 1 000 MW ultra-supercritical boiler desupercritical water system. The results show that the method is efficient, easy to use and stable, which can effectively improve the performance of steam temperature control and achieve good control effects.
