Global dynamic model-based GPC for optimized drum boiler main steam temperature control system

The drum boiler main steam temperature is characterized by complex nonlinear dynamics, so control with conventional PID cascade control is rarely satisfactory. Therefore, an RBF-ARX control model based on the SOM was developed to describe the main steam temperature dynamics with its parameters identified offline from field data. A generalized predictive control (GPC) method with recursive optimization and feedback was then developed based on the SOM-RBF-ARX model. Simulation of a 220 t/h drum boiler and tests on a 130 t/h drum boiler in the field show that the model accurately predicts and tracks variations of the working conditions due to fuel and demand changes.