Research On Robust Iterative Learning Control Applied To Batch Process

In the rapid development of manufacturing industry and the high demand of people’sneeds, batch processes which are including many fields in industry factory, such as, the robotmanipulators, chemical reaction, pharmaceutical, rapid treatment system, injection moldingmachine etc, have played great role in today’s diversity life. Their common feature is that theyrun periodically and have a high value-added characteristic. Batch processes generally showthe strong nonlinearity and they usually have no stable working point which makes itdifficulty to design a controller. Because traditional control methods have lower controlperformance and have longer adjustment time, it’s essential for engineers to use the advancedcontrol methods to improve the economic benefits and the quality of the batch process withthe repeat characterize of batch process.Iterative learning control (ILC) is one of advanced control theory which can improvedrepetitive system’s dynamic performance. With previous batch’s errors are introduced tomodify the current batch control signal for improving the tracking performance of the system.So the paper uses iterative learning control technique to track the reference trajectory of batchprocesses, expecting to reach the requirement of batch processes product quality. But theparameter perturbation, external disturbance always happens in batch process system,therefore it requires a combination of robust control ideas to ensure the stability of thecontroller. The paper mainly includes the following aspects:(1) In order to track a given trajectory for batch process with parameters perturbations, arobust iterative learning control is used to track the trajectory. Before designing the controllerwe should know the model of batch process, the paper uses process data to get the timevarying perturbation model. However, the parameter of batch processes change inevitably. Ifthe controller contains the parameter uncertainty, the design of the controller has a robustcharacteristic and can track the reference trajectory at last. With the batch runs, batch processwill get more process data, and we can use this information to update the system model tospeed up convergence of the batch process.(2) In order to track the pressure of the injection molding machine, optimal iterativelearning control and model predictive control are used respectively to analyze their trackingperformance. Injection molding model can be obtained by input and output data identification.Then optimal iterative learning control algorithm is used to track a given trajectory in thismodel.(3) Iterative learning control and model predictive control based on2D Rosser model isused in nonlinear batch reaction to track the given reference trajectory. Turn the dynamicmodel of the nonlinear affine batch reactor into a general2D Rosser model. Then we designof the controller together with iterative learning control and predictive control based on2DRosser model. Even if the simplified system model in the case of larger errors, it is able totrack a given reference trajectory. Predictive control has the ability in dealing with input andoutput constraint, therefore, the constraint system considered into the framework of thecontroller design. In order to verify the feasibility of the algorithm, the proposed control algorithms were applied in the presence of repeat disturbances and random perturbations ofbatch reactor.Finally, summarize is given and point out the iterative learning control in batch processesfor a further research direction. If we consider the iterative learning control applied to thepractice, some practical problems needed to be considered.