Parameter Tuning Of Model-free Adaptive Controller And Its Application

With the quick development of science and technology,automatic equipments have been increasingly applied in various fields.The progress of the industrial production processes gradually go towards precision and complexity,increasing the difficulty of the accurate modeling for the production processes,and challenging the model-based modern control methods.Under this situation,it would be more difficult to meet with the stricter requirement of control performance.Facing the challenges brought by the modeling of the model-based control methods,the data-driven control methods have been attracted significant attention.The model free adaptive control(MFAC),as one of the data-driven control methods,does not demand the model information of the controlled plants,and only use the input/output(Input/Output)data of the controlled plants.Besides,MFAC,with the features of simple structure,fast calculation and wide applicability,has been becoming an important branch of the data-driven control theory.Until now,there exists no systematical parameters tuning methods in the MFAC controller design process,which limits the practical applications of MFAC.Therefore,it is meaningful to study the parameters tuning methods for MFAC for further promotion of MFAC in theory and practical applications.This thesis focuses on the parameters tuning methods of MFAC controller based on full form dynamic linearization(FFDL)for a class of discrete-time nonlinear systems,and its application in the water level control system.The details are described as follows:(1)Two kinds of parameters tuning methods for the MFAC controller are proposed,including the methods based on the phase margin and the methods based on the optimal performance index.Firstly,by using the FFDL technique,a general discrete nonlinear system can be transformed into an equivalent dynamic linearized data model in real time,based on which the MFAC controller is designed.Further,the designed MFAC controller is converted into the structure like incremental PID controller.Based on this structure,two kinds of PID parameters tuning methods are applied to FFDL-MFAC controller to tune the parameter step factors of the cotnroller.Finally,the effectiveness of two proposed methods is verified by numerical simulations.(2)Under the framework of FFDL-MFAC controller,a pseudo gradient tuning scheme based on virtual reference feedback tuning(VRFT)is proposed.VRFT can be utilized to parameters optimization suning,and by using this advantages of VRFT,the initial values and reset values of the pseudo gradient in FFDL-MFAC controller is tuned,solving the difficulty of determining the reset values when the pseudo-gradient initial value selection and the system structure in the FFDL-MFAC controller change.Finally,the effectiveness of proposed scheme is verified by simulations.(3)In order to further verify the validity of the abovementioned two types of parameter tuning methods,a water tank level system is adopted for simulation analysis.For comparison with the PID-based FFDL-MFAC and VRFT-based FFDL-MFAC tuning methods,the traditional PID and FFDL-MFAC control methods are used here..Simulation results demonstrate that the effectively tuned parameters of the two proposed PID-based and VRFT-based FFDL-MFAC controllers can be achieved and they yield improved control performance.