Research On Robustness For PVC Reactor Temperature Control Based On Iterative Learning Control

The produce of PVC resin,which has the characteristic of repeat running,is conducted in the intermittent reactor.The mathematic model has difficulty in establishing because the reactor has the Characteristic of nonlinearity,uncertainty, and external interference,and the conventional control methods do not meet the industry control.Therefore,a kind of advanced control strategies suitable for the condition is urgently required.Iterative learning control is a new intelligent control approach,which is developed in the recent 20 years.And this control approach is suitable for the occasion with the characteristic of repeat running.It does not depend on the precise mathematic model of dynamic systems,but to utilize the prior control experience of the control system to search an ideal input signal on the basis of the practical output signal and the expected output signal.It is an optimized input signal generated from iteration,and it can reduce the error by executing the same task repeatedly and enable the system output to approach the ideal values.It has a very important significance for the control problems with nonlinearity,complexity, difficult modeling and high-precise track.Because open loop iterative learning control algorithm only utilizes the former running information of the system,it has on robustness to the non-repeatable interference,and has no equanimity effect to the controlled object. As a result,,it has the probability to generate large tracking error even as the learning rule meets the convergent condition during the learning process.In contrast,the closed loop iterative learning control algorithm only utilizes the current running information of the system,and it can track the expected track precisely only if the feedback plus is very large.But the saturation of the executer and so on makes the high plus feedback meaningless in the realization process. Therefore,both of them have some weaknesses from the view of the use of the control information.For the reasons above,in this thesis we introduce a PID feedback controller to construct feedback-forwardback iterative learning control on the basis of open loop PID iterative learning controller,aiming at the temperature control of the PVC polymerization reactor,in order to improve the tracking and robustness performance of the iterative learning control.This controller is composed of feedback controller and forward-back controller,and use the rule coupled by open loop and closed loop,that is to say,it synthesizes to utilize the current and former running information of the system to correct the learning control strategy of the current control input of the controlled object.The feedback controller is mainly used to realize equanimity task,and makes the system output no far from the expected track,and ensure the system stable,and fasten the convergent speed.Simultaneously,it can use the feedback energy to constrain the interference and improve the learning robustness.In the thesis,the robust control of open-closed loop iterative learning control is extended to open-closed PID.As a result,it can adjust the proportional plus, integral plus and differential plus to overcome the system disturbance simultaneously,and makes the adjustive measures flexible,and improve the tracking performance of the system.The research on the robustness of the disturbed system is conducted on the control of open-closed loop PID iterative learning control rule.And the robustness condition of open-closed loop PID iterative learning control of the reactor is drawn.Under the condition of the existence of uncertainty,the shift of the initial state,the state inference and the limitary output,we prove that the system output is bound convergence,and the boundaries have relationship with the boundary of the inference of the two bordered iterative running,and the system output is consistently convergent to the expected track if these inferences repeat asymptotically.The final simulation results show the validity of the approach.