焦炉烟气脱硝系统的阶梯式广义预测控制

Stepped Generalized Predictive Control for Coke Oven Flue Gas Denitrification System

近年来,随着GB16171—2012《炼焦化学工业污染物排放标准》的制定,我国的焦化企业相继建立了焦炉烟气脱硝设施,但该设施在被控过程中呈现出高延时和大惯性的特点,无法以传统的PID控制方法实现令人满意的控制目标。为破解这一难题,提出了一种基于阶梯式广义预测控制的焦炉烟气脱硝的方法。基于工业现场采集的数据,建立了被控对象的数学模型;使用CARIMA模型描述被控对象,结合渐消记忆递推最小二乘辨识,得到模型参数;根据被控对象的数学模型设计了阶梯式广义预测控制器,并导出控制器的闭环特征多项式,从而证明了控制器的闭环稳定性;利用Matlab软件建立了焦炉烟气脱硝控制系统的仿真环境,并对上述控制器开展了模拟测试。仿真结果表明,相较于传统PID控制器,阶梯式广义预测控制器具有更好的鲁棒性能,对深入推进和落实焦化企业焦炉烟尘超低排放技术改造具有深远意义。

In recent years,with the formulation of GB16171-2012 Emission Standard of Pollutants for the Chemical Industry of Coking,coking enterprises in China have successively established coke oven flue gas denitration facilities.However,the facilities are characterized by high delay and large inertia during the controlled process,and can not achieve satisfactory control objectives with the traditional PID control method.To solve this problem,a method of coke oven flue gas denitrification based on stepped generalized predictive control is proposed.Based on the data collected in the industrial field,the mathematical model of the controlled object is established.The CARIMA model is used to de-scribe the controlled object,and the model parameters are obtained by combining with the recursive least squares identi-fication of fading memory.According to the mathematical model of the controlled object,a stepped generalized predictive controller is designed,and the closed-loop characteristic polynomial of the controller is derived,which proves the closed-loop stability of the controller.The simulation environment of coke oven flue gas denitration control system is established by using Matlab software,and the simulation test of the above controller is carried out.The simulation results show that the stepped generalized predictive controller has better robustness than the traditional PID contrller.It is of far-reaching significance to further promote and implement the technical transformation of ultra-low emission of coke oven smoke and dust in coking enterprises.