A Boiler Main Steam Temperature Control Based On Muti-model Prediction

The boiler main steam temperature control has been a difficult issue of industrial control for a long time and the main reasons have two aspects. Firstly, there exist the large delays and inertias between the main steam boiler's control quantity and its output (the main steam temperature). When taking the corresponding control to the change of the output, it is actually impossible for the control quantity to play a role on the detected change and thus the control effect is quite poor due to the existence of these large delays and inertias. In addition, the system parameters alter as the boiler loads change. For a boiler temperature model, it is not easy to build an accurate mathematical model because the parameters of the controlled objects are under alterations, which lead to the difficulty in establishing the fixed parameter model to realize the accurate control. Thus, on the basis of the two reasons above, boiler control is still a difficult issue during the modern production process.When the boiler has a large parameter changes, the effect of a typical PID control plays not as well as an experienced teacher's manual control. The reason is that under the manual control, people will forecast a control model according to the current load situations and other conditions and then they are able to have a further control in terms of their own control experience. Therefore, if the control algorithm had an intelligent performance like human control, it would greatly improve the control effect. At present, a lot of researches on the boiler control algorithm are also based on that. The popular ones are the predictive control and fuzzy control.This paper proposes the predictive multi-model switching based on the predictive control. On the one hand, we can forecast a future model according to the current output. On the other hand, setting several kinds of typical model for switching when forecasting can not only overcome the influence from the control object's large inertias, but also the one from its parameter change. Through the simulation results under different load situations in Simulink, we obtain the pros and cons between the multi-model predictive and PID control, which draws the conclusion that the control effect of the former is better than that of the latter. We also have an appropriate analysis to the multi-model switching through the simulation and provide the corresponding comparisons with the effect of predictive control when the disturbances exist in system.In terms of the application, this paper does not spend lots of space on a deep discussion about the prediction control algorithm and multi-model switching, but on the simplification of two control methods for the main steam temperature control system. On the one hand, the prediction is mainly to take the generalized predictive control and the actual data to establish a prediction model in advance. On the other hand, the switching index of multi-model switching is based on the optimal index from the generalized predictive control. Finally, we also achieve the programm of the multi-model predictive control algorithm by the S-function.