Application Study Of Multivariable Generalized Predictive Control In Boiler Combustion System

The boiler combustion control system is a complicated multiple input multiple output system, the characteristics of nonlinear and large time delays make the conventional PID algorithm not well satisfy their process requirements. As a kind of algorithm which developed from industrial production process, generalized predictive control has a very good control effect for solving such problems. In this paper, the multivariable generalized predictive control algorithm is applied in boiler combustion system, the good simulation results are obtained.The main contents of the research in this paper are as follows.(1) Under the premise of understanding the process of combustion system and the requirements of control system,we put forward the specific control scheme under different conditions of each link, then the difficulties of boiler combustion control system is analyzed. Because the effect of dealing with complex industrial process with PID control algorithm is not satisfactory, this paper presents a new method that the multivariable generalized predictive control algorithm is applied in the system, for this algorithm is suitable for the system with uncertainties object and large time-delay, strong coupling, multivariable characteristics.(2) This paper study on the predictive control algorithm. Then, this chapter introduces the theory of generalized predictive control algorithm from both SISO (single input and single output) and MIMO (multiple input and multiple output) system. But the traditional generalized predictive control has one important problem which is the large number of online computing. So the improved generalized predictive control is presented. Then the simulation reveal that the algorithm is feasible.(3) We know there are some problems existing in the boiler combustion system which uses the traditional PID control. The improved multivariable generalized predictive control algorithm is applied to the boiler combustion system in the power plant. The results of simulation tell us the feasibility of the algorithm.