基于扩张状态观测器的双母管系统预测控制

Predictive Control of Double Main-pipeline System Based on Extended State Observer

双母管式机组较为广泛地应用于热电联产机组中,但由于多炉多机和2根大容量母管互相影响,导致热电负荷跟踪不及时,母管压力控制自动化水平较低。为此,针对双母管系统的非线性、强耦合、大迟延特性,设计了基于广义扩张状态观测器的多模型预测控制(generalized extended state observer based muti-model predictive control,GESOMMPC)方法。首先,建立了基于间隙度量(gap-metric)的多模型控制对象用于逼近非线性系统;其次,设计了扩张状态观测器估计系统耦合的集总扰动,并作为前馈信号输入到预测控制器中;最后,设计基于扰动前馈的多模型预测控制器实现对双母管系统的控制。实验结果表明,相对于PID方法,所提方法在满足电热负荷的同时,可以在允许范围内保持母管压力稳定,且动态偏差更小,过渡过程时间更短。

The double main-pipeline unit is widely used in the cogeneration unit.However,due to the mutual influence among multiple furnaces,multiple turbines and the two large-capacity main-pipeline,thermoelectric load is tracking untimely and the automation level of the pressure control of the main-pipe is low.To improve nonlinearity,strong coupling,and large delay characteristics,a multi-model predictive control method based on the generalized extended state observer(GESOMMPC)is designed.Firstly,a multi-model control object based on the gap metric(gap-metric)is established to approximate the nonlinear system;Secondly,an extended state observer is designed to estimate the lumped disturbance of the system coupling,and input to the predictive control as a feedforward;Finally,a multi-model predictive controller based on disturbance feedforward is designed to control the double-main-pipeline system.Compared with PID,the experimental results show that the scheme can keep the pressure of the main-pipe within the allowable range while meeting the electric heating load,and controlled variables get smaller dynamic deviation and shorter transition process time.