Performance Assessment For Iterative Learning Control Based On 2-D Fornasini-marchesini Model

With the rapid development of industrial process,the control system in industrial process tends to be large-scale and complex,and the performance of control system is widely concerned by the society.As an efficient tool to evaluate the performance of the control system,control performance assessment shows the performance of the control system directly.Although it is a relatively young branch of research,it has received extensive attention from both(control)industry and(control)academia.As one of the effective tools to deal with the industrial batch processes,the research on iterative learning control(ILC)has been developed rapidly since it was first put forward,and a large number of advanced algorithms have been proposed.However,the study and analysis of performance assessment on ILC system is neglected.At present,there are only a few publications on the performance evaluation of ILC.However,in order to ensure the industrial process performs well and stable for a long time,it is necessary to pay enough attention to the performance of control system.Therefore,performance assessment of ILC is studied in this thesis,which contains the analysis of ILC system based on different models in two situations:model-known case and model-unknown case.The main contents of this study are as follows:1.As a branch of control research field,control performance assessment has obtained much more attention and been researched widely in academia.According to different system characteristics,different assessment benchmarks have been proposed one by one.In this study,the issue of research in this field and related fields are briefly reviewed and summarized.We analyzed the advantages,adaptability,and limitations of different benchmarks,which provided a theoretical basis for the subsequent research.2.In the light of the characteristics of ILC system,the batch processes controlled by ILC were modeled.We analyzed the properties of the transferred two-dimensional(2-D)model,and designed a 2-D LQG benchmark for performance assessment.By using the linear matrix inequality,we solved the controller and established the control performance assessment system.Several simulations verified the directness,feasibility and effectiveness of the proposed method.The performance assessment of ILC based on different 2-D models were analyzed and compared in the same framework,the results were illustrated by the simulations.3.Research on control performance assessment usually relies on the condition that the system is known clearly.When we do not acquire the system parameters,we should turn to system identification technology without disturbing the normal operating system.In this study,we deeply studied the performance assessment under model-unknown case.Subspace identification was applied to the transferred 2-D model in the unified framework,which obtained the system parameters accurately and was familiar with the system deeply.The simulation examples showed the feasibility of the method,and improved the control performance assessment system.