Iterative Learning Control For Batch Processes With Time Delay And Nonrepetitive Disturbances

Batch process is a common production mode in modern process industry,which is widely used in injection molding,drug crystallization,semiconductor production and other processes due to its characteristics of meeting the needs of multiple types and refined production.Because of the material transmission,signal transmission etc,there is a common time delay phenomenon in industrial batch processes.If the control method is designed directly without considering the influence of time delay,the system performance may be reduced or even become unstable.In addition,due to the time-varying characteristics of process or the change of external environ-ment,batch processes also have a wide range of internal uncertainties and external disturbances.How to effectively suppress or eliminate the disturbance is an important problem to be solved in industrial control design and performance optimization.Iterative learning control(ILC)is an intelligent control method,which gradually improves the control performance of the current batch/trial by constantly learning the historical information of the previous batch,and finally realizes the complete tracking of the desired trajectory within a limited time.Although ILC has been proven to be an effective control method for batch processes,traditional ILC often requires the system to have strict repetitive conditions,and there are still many problems to be solved in how to deal with non-repetitive uncertainties and disturbances validly.In this paper,an ILC design method based on output measurement and disturbance compensation is studied for time delay batch processes with non-repetitive uncertainties and disturbances,especially when the actual state information is difficult to be completely measured.The main research contents and innovations include:1.For discrete batch processes with time-varying state delays and non-repetitive uncertain-ties,an output feedback based PD-type ILC scheme is designed.The proposed method only uses the output information,which is convenient for practical industrial process application.Based on the linear repetition process model and the Lyapunov-Krasovskii stability analysis theory,sufficient conditions are established in terms of linear matrix inequalities(LMI)to ensure the system is robustly stable along the trial.In order to avoid the problem of non-convex stability conditions caused by output feedback,a state feedback based PD-type ILC method is designed and a two-stage heuristic algorithm is proposed to obtain the output feedback controller.The algorithm clarifies the connection between output feedback and state feedback,and can ensure the stability condition is feasible by iterative solution.Finally,an injection molding simulation is used to verify the effectiveness of the proposed method.2.An equivalent input disturbance(EID)based ILC method is proposed for time-varying batch processes with non-repetitive uncertainties and external disturbances.Firstly,the influ-ence of internal uncertainties and external disturbances on the output are uniformly equivalent to the existence of an EID with the same effect on the input.Then,an EID estimator is de-signed to estimate the disturbance,including a full-order state observer and a low-pass filter.An ILC design method with feedforward disturbance compensation is proposed by connecting the filtered EID estimate to the input terminal.Based on Lyapunov-Krasovskii stability analysis theory,sufficient LMI conditions to ensure robust stability of the process along the trial were derived.The observer and ILC controller gains can be obtained by solving the established LMI.Finally,the effectiveness and advantages of the proposed method are proved by an experimental simulation of injection molding.3.For a class of batch processes with time varying uncertainties and disturbances,an indirect-type ILC scheme based on extended state observer(ESO)is proposed.Different from the common direct-type ILC design,the indirect-type ILC scheme consists of an inner loop and an outer loop,and the two parts can be designed independently by using the separation prin-ciple.In the inner loop,the total disturbance is defined first,and the ESO is constructed to estimate the state and total disturbance of the system.Based on the estimated state and distur-bance information,an inner loop feedback control law is designed to ensure the robust stability and disturbance rejection performance of the system.In the outer part,a simple P-type ILC is designed to improve the tracking performance along the trial direction.Finally,an illustra-tive example of injection molding is conducted to verify the validity and merit of the proposed scheme.