针对火电厂选择性催化还原(Selective Catalytic Reduction，SCR)烟气脱硝系统机理复杂，工况变化时呈现的不确定性、强扰动等特点，提出了一种基于互信息和PID神经网络的SCR烟气脱硝扰动补偿控制方法。利用PID前向神经网络的学习性能逼近被控对象的逆构成扰动观测器对系统进行反馈补偿，以达到超前消除系统扰动的目的。选取观测扰动和系统扰动的互信息为目标函数，采用改进的帝国竞争算法实现PID神经网络权值的优化调整。设计鲁棒PID控制器来进一步克服被控对象存在的不确定性。仿真实验表明，该方法具有突出的抗干扰能力和较好的鲁棒性，控制品质优于常规的PID控制。

The Selective catalytic reduction (SCR) denitration system is very complex and has the characteristics of strong disturbance and uncertainty.Based on mutual information and PID neural network,a disturbance compensation control method for SCR denitration system was proposed in this paper.The disturbance observer was constructed by using the PID feedforward neural network to approximate the inverse of the controlled plant.The disturbance observer was used to compensate the control system in order to eliminate the disturbance in advance.The mutual information between observation disturbance and system disturbance was selected as the objective function,and the improved imperialist competitive algorithm was used to optimize the parameters of PID neural network.The robust PID controller was designed to overcome the uncertainty of the controlled plant.The simulation results show that the method has outstanding disturbance rejection performance and good robustness,and the control performance is better than that of the general PID control.

TP273

海南省自然科学基金（619MS070）;2020年度海南热带海洋学院科研启动资助项目（RHDRC202004）