Application Of Multivariable Generalized Predictive Control Algorithm To Monoblock Unit Coordinate Control System

In this thesis, a centralized optimization direct multivariable generalized predictive control algorithm has been proposed, based on the principle of generalized predictive control. On one hand, this method requires lowly for the description of model and has high robustness, as same as the fundamental GPC algorithm. On the other hand, because it reduces the computational load a lot by using the implicit model of the plant, the method is easy to be realized on-line. Moreover, because this method radically overcomes the weakness and difficulty of decoupling method, it improves the reliability and control quality of the system greatly. On the basis of theoretic analysis of the method , this thesis studies its application to monoblock unit coordinate control system by dynamic simulation, considering the characteristic of capacious unit. The control results of the algorithm and the other two are compared in different operational conditions. Simulation shows that monoblock unit can be controlled well by using this method. Therefore, it not only has important theoretical value but also has high practical value to study the application of direct MGPC method in thermal control system.