针对选择性催化还原(SCR)烟气脱硝系统具有非线性、大惯性、大延迟、多干扰的特点,造成出口NOx浓度无法达到理想设定值、喷氨过量等问题,本文提出将自适应广义预测控制器(GPC)应用于SCR烟气脱硝控制系统。使用可变遗忘因子的递推最小二乘(VFFRLS)法,根据从历史运行数据中获取的有效动态过程信息辨识得到脱硝系统控制对象不同工况区间的过程模型,并在此基础上进一步研究了基于GPC算法与VFFRLS在线辨识算法的自适应GPC在脱硝系统中的应用效果。仿真对比和扰动测试及变工况实验表明,相比于普通GPC与串级PID控制,带有VFFRLS辨识模块的自适应GPC调节速度更快、鲁棒性更强,具有良好的抗干扰能力和适应能力。

Selective catalytic reduction (SCR) denitrification system has the characteristics of non-linearity, large inertia, large delay and multi-disturbance. These characteristics lead to the problem that the concentration of NOx emission can not reach the ideal set value and ammonia injection is excessive. In this paper, we propose that the adaptive generalized predictive controller (GPC) is applied to SCR flue gas denitrification control system. By using the recursive least squares (VFFRLS) method with variable forgetting factor, and based on the identification of the effective dynamic process information obtained from the historical operation data, the process model of denitrification system control object in different working conditions is obtained. On this basis, the application effect of adaptive GPC based on GPC algorithm and VFFRLS online identification algorithm in denitrification system is further studied. The simulation comparison, disturbance test and variable working conditions experiments show that compared with conventional GPC and cascade PID control, the adaptive GPC with VFFRLS identification module has faster adjustment speed, stronger robustness, good anti-interference ability and adaptability.

TK39

上海市科委地方能力建设项目（18020500900）；上海市自然科学基金（19ZR1420700）