Design Of Nonlinear State Observer And Its Application In Supercritical Unit

Compared with subcritical unit,supercritical unit is more competitive in the power market because it is cleaner,greener and more efficient.However,its dynamic characteristics are complex and supercritical unit is difficult to control to meet the high control requirements.Therefore,it is very necessary to research new control strategies in supercritical unit to improve its control quality.This paper studies the design and improvement of an extended Luenberger state observer in order to compensate uncertain factors of supercritical units and improve supercritical units' complex dynamic characteristics.In addition,this paper introduces a new control strategy with predictive control in order to handle multiple variables and large time delay of supercritical units.First,this paper researches an extended Luenberger state observer according to the design principle of ESO in ADRC.The observer not only can estimate system states but also can compensate system uncertainties.However,parameter setting of the designed observer is complex and there is chattering phenomenon when the error of initial observer states and initial system values is large.Therefore,the sigmoid excitation function in neural network is improved to build an improved extended Luenberger state observer.The improved observer dose not change the original capacities of observation and overcoming system uncertainties and it can reduce system chattering phenomenon.Finally,a novel control strategy is designed based on the combination of the state feedback with the improved extended Luenberger state observer and predictive control algorithm.In this paper,two researched nonlinear state observers and the new control strategy are all simulated in supercritical unit to verify their feasibility and effectiveness.Simulation results show that the two observers are good at observing system states and the improved observer can effectively inhibit flutter thus improving system security and stability.In addition,compared to alone predictive control,the novel strategy including the state feedback has better rapidity,accuracy and decoupling effect.