抄纸过程定量水分系统的预测控制

Predictive Control of Quantification and Moisture System in Paper Process

由于抄纸水分控制系统具有非线性、大滞后、强耦合、快时变等复杂特性,造成抄纸过程定量水分系统控制性能差。提出了单神经元预测控制算法,将神经元PID控制及预测控制的优点相结合,通过在传统PID控制中引入单神经元控制,实现了对PID参数的在线调整,提高系统的自适应性和鲁棒性;通过引入预测控制实现了控制器参数提前调整,加快了调节过程。上述算法用单神经元PID对预测误差进行补偿,弥补了传统预测控制对预测信息处理的不足。将改进控制算法与Smith预估控制及动态矩阵控制(DMC)相比较,仿真结果表明,改进控制算法效果较好。

A single - neuron predictive control algorithm is proposed for quantification and moisture system in pa- per process, this system has complex features such as nonlinear, large time delay, strong coupling, fast time - varying and so on. The algorithm combines the advantages of single - neuron PID control and predictive control. The single - neuron control is put in the traditional PID control to realize the online adjustment of PID parameters, improve the adaptability and robustness of the system. The predictive control realizes the adjustment of controller parameters in advance, and speeds up the adjustment process. The single - neuron PID controller is used to compensate the possible prediction error. Simulation results show that, compared with the Smith prediction control and DMC, the proposed algorithm has better control effect.