Main-steam temperature control for ultra-supercritical unit using Multi-Model Predictive strategy

Abstract

A Multi-Model Predictive Control (MMPC) strategy based on dynamic matrix algorithm is proposed and applied to the main-steam temperature control of a ultra-supercritical once through boiler-turbine system in this paper. Firstly, models and corresponding controllers can change with the changing operating point via a multi-model switching technique so as to achieve robustness. Secondly, by multi-step prediction, rolling optimization and feedback correction, the plant output is optimized at each sampling interval so as to obtain better dynamic performance. Thirdly, due to good real-time tracking performance, the system can respond faster. Furthermore, in order to inhibit the sudden disturbance, a inner loop of proportional is added to form a cascade MMPC-P controller. Simulation shows much better robustness and dynamic performance for various kinds of electric load demand changes and parameters variations via this strategy than the conventional PID method.