Study On Feedback-assisted PD-type Quantized Iterative Learning Control Methods

Iterative learning control(ILC)is inspired by the human learning process with the key idea that learning from repetition,thereby improving the transient response and tracking performance of the systems.To improve the robustness under non-repetitive interference,iterative learning control and feedback control have combined with complementary advantages.It forms a feedback-assisted iterative learning control method,which enhances its practicality.With the development of network communication technology,iterative learning control theory is gradually applied in Networked Control Systems(NCSs).Quantization is one of the effective methods to reduce the network transmission load and improve the system operation efficiency.The research on iterative learning control is still in its infancy.In this paper,the convergence and robustness of feedback-assisted PD-type quantized iterative learning control are studied.P-type and PD-type learning laws are given at the same time,and the superiority of the proposed algorithm in accelerating the convergence speed is verified by comparison.Specific work includes:1.Combining the ideas of quantization and feedback-assisted iterative learning control,a feedback-assisted PD-type quantized iterative learning control scheme is proposed,and the design and convergence analysis of the system controller when quantizing the error signal are discussed preliminarily.In the following specific work,we will expand the discussion of the three cases of packet loss,random variable trial length and variable initial conditions to solve the problem of system data loss and non-repetitive conditions.2.The problem of iterative learning control under communication constraints is studied.During the transmission of the output signal from the remote device to the controller,there are data packet loss and data quantization.The packet loss model is described as a Bernoulli sequence.A new feedback-assisted PD-type quantized iterative learning control scheme is designed,and the compression mapping method is used to prove that the proposed algorithm can ensure that the tracking error converges to zero.Then,the robustness of the algorithm is discussed for the initial state shift.3.The iterative learning control problem in which the trial length and initial state change randomly is studied.Based on the analysis of the error quantized signal,the scheme of the input quantized signal is added.Based on the corrected tracking error,a new feedback-assisted PD-type quantized iterative learning control algorithm is designed,which relaxes the requirements of trial length and initial conditions for repetitive systems in classic iterative learning control.Using the lifting system framework,the analysis proves the learning convergence and robustness of the two quantization schemes under mathematical expectations.