The Research Of Marine Boiler Control Algorithm Based On Model Predictive Control

The control of marine boiler is a complicated nonlinear time varying control; there is a strong coupling phenomenon in boiler operation parameters (such as water temperature, steam pressure, water flow, water level, etc.) The traditional marine boiler is used the feedback control of PID controller, but this kind of control is easy to produce overshoot, bad robustness. For this complex system, this paper chooses model predictive control in order to obtain a more ideal control effect.This paper first has a brief discussion of Aalborg industries A/S MISSIONTM OB marine boiler, and the dynamics of total mass and energy balances is considered, then the expression of water level and steam pressure mathematical equations is got. The whole article is based on the established mechanism model here. The next, the paper discusses the development, principle, and basic characteristic of model predictive control. Then the paper presents a MPC algorithm in the application of this boiler, and a simulation study is made. And this paper designs a state estimator based on kalman filtering theory before the application of MPC, in order to get the state variables that cannot be directly measured in state space equation. To realize the simulation results, the paper put the linear MPC as the key part. That is the first the article must linearize and discrete the mechanism model, and establish linear discrete state-space model, so that the algorithm can be simulated directly using Matlab. This paper has a simulation analysis with various steam load disturbance.By comparing the resulting simulation dynamic drawing of MPC controller and PID controller, the paper can clearly prove that the control performance of MPC controller is far better than the PID controller with various steam load disturbance. In other case, the MPC controller can effectively reflect the actual dynamic boiler system and have better control effect, that is, without overshoot, stability required in a short time and good performance in tracking desired value.