The Study Of Paper And Pulp Making Based On Single Neuron PID Control With Variable Gain

In the new normal of our economy,pulp and paper industry has entered a new phase of reform.The new stage requires the papermaking industry to eliminate backward production capacity and to reduce cost & save energy.In addition,the orientation and focus for papermaking industry in the new stage are not quantity but quality,upper lever and higher grade,which means control performance requirements on papermaking industry are more and more strict.At the same time,the progress of paper and pulp making is very complex,with nonlinear,uncertainty,timevarying,large time delay,large coupling,large inertia and other characteristics.And conventional PID control needs to rely on accurate mathematical model,therefore it cannot satisfy the higher control performance requirements.With the development of computer technology and control theory,advanced control has also been increasingly applied in the process of pulp and paper industry control.Neural network control is a kind of advanced control.Its nonlinear fitting ability and selflearning ability is far superior to conventional PID control,but it has a slow convergence rate and its structure of the selection is still no regular,it is also easy to fall into Local minimum.Single neuron PID control combines the advantages of neural network control and conventional PID control.Thus,single neuron PID control has the self-learning and adaptive ability,simple structure,strong robustness.But its gain K cannot be adaptive.Therefore,to study the variable gain of single neuron PID control in the process of pulp and paper making has a significance.Typical processing points in the three sections of pulp and paper making(pulping stage,paper making stage and alkali recovery stage)are selected as the controlled objects.In order to verify the effectiveness of the variable gain of single neuron PID control,the controlled objects are replacement temperature difference of pulping stage,pulp consistency of papermaking stage,causticizing temperature and black liquor level of alkali recovery stage.A single neuron PID algorithm based on expert experience,single neuron PSD algorithm,single neuron PID algorithm based on immune mechanism and fuzzy-single neuron PID algorithm are designed respectively.The simulation experiments are carried out in Matlab,where calling s function to achieve the algorithm function.The results show the improved single neuron PID algorithm has faster response speed than the single neuron PID algorithm,and has stronger anti-interference,better dynamic and static performance than the conventional PID algorithm.To verify the algorithm,selected one of the variable gain algorithm-single neuron PID algorithm based on expert experience and apply it to the nonlinear liquid level system of experimental platform.The experimental results show that the algorithm has better dynamic and static performance,and has the feasibility compared to the conventional PID algorithm.