Design And Evaluation Of Batch Process Controller Based On Iterative Learning Control

The disturbance attenuation in batch process has been a challenging issue for control field.In actual industrial control process,the products quality may deteriorate as a result of environmental noise,or other uncertain disturbance.It is difficult to eliminate these disturbance by simple controller design or filtering.The controlled trajectory could deviate from the desired one if it is not properly controlled.Therefore,in this thesis an integrated method combined system filter with controller design is proposed to solve stochastic disturbance in iterative learning control system.Main work includes:1.The detrimental effect of stochastic disturbance in iterative learning control system is studied.The system performance with stochastic disturbance was firstly discussed and the controller was designed which can achieve a certain anti-jamming effect of a given batch iteration.However,with the iteration number increased,the random noise during each batch caused a large amount of accumulation of system,which decrease the control performance.To allievate such a problem,a preprocessing of the output error was introduced the two repeated experiments based on wavelet transform filter was used to filter the output error,which led a better performance.2.A design algorithm of iterative learning controller based on the liner Infinite Impulse Response(IIR)digital filter using least squares identification was proposed.In view of controller design,since the wavelet transform has no analytical expression,which leads the recursive expression of stochastic error parts cannot be obtained for the controller design.This IIR filter has the equivalent filtering effect of the original wavelet filter.The stochastic error parts were filtered by IIR filter with its recursive expression obtained simultaneously.Thus the optimization process can be converted into a convex optimization problem with minimum the objective function for the controller design.Experiments show that the proposed algorithm achieves good anti-jamming performance,and the accumulation of random noise was reduced between batches.Finally,to illustrate the validity of the proposed algorithm,the controller performance was evaluated based on the minimum variance criterion.