As the superheated steam temperature of thermal power units has characteristics of great inertia

pure time delay and strong disturbance

the generalized predictive control strategy was adopted into cascade temperature control system. Through rolling optimization of objective function

estimating and correcting the model of controlled object at each sampling time

and using recursive algorithm to reduce online computation work

the computation time was reduced and real-time of the algorithm was enhanced. The simulation comparison results prove that

with low parameter drift

stability and better economy

the proposed algorithm has good control performance in superheated steam temperature control.