Predictor-based Sampling Control Of Time Delay Processes &batch Optimization

In this thesis,firstly,on the basis of GP structure which can draw delay out of the closed loop,a two degrees of freedom control algorithm is proposed for processes with dead time.Based on the IMC theory and robust control theory,the disturbance rejection controller is obtained by reverse the expected complementary sensitivity function of the closed-loop system and the set-point tracking response controller is designed according to H2 optimal control theory.The merit of the proposed controller design method is that the set-point tracking response of the system and load disturbance responses can be independently adjusted and optimized.In addition,the index of rising time can be adjusted by a single control parameter to realize the quantitative analysis.For a particular type of model mismatch,the thesis gives a graphical method for the tuning of the parameter in the anti-jamming controller.Simulation cases and temperature control experiments show the superiority of this method.To simplify the control structure and modify the control performance,a simplified generalized predictor is proposed.Compared with GP structure,the external loop is reduced and the predictors used to estimate the un-delayed output are modified by adding a filter.The prediction performance and robust stability can be tuned by a single parameter in the filter.The merit of the SGP is that the control structure is simplified such that the system stability is enhanced while ensuring the un-delayed output.Simulation cases and temperature control experiments show the superiority of the SGP control structure.On the basis of the SGP structure,a scheme in terms of iterative learning control(ILC)is proposed and the main point is to use the SGP structure to construct the un-delayed output and then obtain the estimated state by the estimator.Combined with the 2-dimensional robust predictive control based iterative learning control scheme,the controllers can be solved by the linear matrix inequalities using the un-delayed feedback.The merit of the scheme is that the delay can be canceled in the feedback loop.Besides,based on the 2-dimensional robust predictive control,the 2-dimensional historical information can be used adequately to ensure the convergence of iteration error along the iterative index.Finally,an example is used to demonstrate the superiority of the control scheme.