Quantized Iterative Learning Control And Optimization Based On Information Coding

Iterative learning control(ILC)is an intelligent control method that is suitable for performing repetitive tasks.By learning the historical information of the completed trials to update the control signal of the current trial,the operation performance of the system is gradually improved.With the development of communication technology,the research on iterative learning control in the network environment is becoming more and more extensive,and quantization is an effective method to reduce the load of network communication and wireless transmission and improve the operating efficiency of the system.The combination of iterative learning control and quantization to solve the trajectory tracking problem of network systems or wireless control systems with repetitive operation characteristics has important theoretical and practical significance.In this paper,the research on quantized iterative learning control and its performance optimization algorithm is carried out based on information encoding,which refers to the process of encoding and decoding the signal.The main research content of this thesis is summarized as follows:(1)Aiming to point-to-point tracking problem of discrete linear time-invariant systems with input signal quantization,the input signal is quantized by a combination of an infinite logarithmic quantizer and an encoding-decoding mechanism.Thus the adverse effect of excessive quantization error on the tracking performance in the direct quantization input scheme is avoided.The performance index function is constructed under the norm optimization framework and approximated to obtain the corresponding point-to-point quantized iterative learning control update law,which proves that the system can achieve tracking error convergence by using the update law.Finally,the proposed control law is applied to the DC motor model,and the simulation results verify the feasibility and effectiveness of the method.(2)The design of iterative learning control algorithm combining infinite logarithm quantizer and finite logarithm quantizer with encoding and decoding mechanism is studied.Starting from the case of finite logarithmic quantizer,under the framework of norm optimization,the boundedness of relative quantization error and the properties of norm are used to construct a new performance index function,and then the corresponding quantization iterative learning update law is designed to realize the input signal point-to-point tracking targets for quantification systems.Then it is extended to the more practical case of finite logarithmic quantizer,and the value scheme of saturated quantization value is given.The proposed control law is applied to the DC motor model,and the simulation results verify the feasibility of the method.(3)Aiming at the full trajectory tracking problem of the input and output signal quantization system,the input end uses the coding and decoding quantization scheme,and the output end uses the direct quantization tracking error signal scheme.On this basis,the robustness of the designed update law under actuator failure is further studied,and it is proved that with the increase of the number of iterations,the system output can re-track the reference trajectory after the actuator failure.Finally,the effectiveness of the proposed method is verified by the simulation of the permanent magnet linear motor model.