Zero‐error convergence of iterative learning control based on uniform quantisation with encoding and decoding mechanism

Abstract

In this study, the zero-error convergence of the iterative learning control for a tracking problem is realised by incorporating a uniform quantiser with an encoding and decoding mechanism. Under this scheme, the system output is first transformed and encoded. Then, the encoded information is transmitted back for updating the input. The results are extended to a finite quantisation level situation under the same framework and a simulation using a permanent magnet linear motor is performed to demonstrate the effectiveness of the proposed scheme.

Model‐free adaptive control for a class of nonlinear systems with uniform quantizer

“…Here, we consider the encoding-decoding mechanism in References [35][36][37] and they are applied to MFAC. For the system measurement output data y(k), the associated encoder is designed as…”

“…(3) This article has investigated the problem of data quantization for a class of SISO unknown model nonlinear discrete systems. However, the existing quantized control design with encoding and decoding mechanism in References [35][36][37] only considers linear or nonlinear systems with the known model information. It should be pointed out that the introduction of encoding and decoding mechanism for MFAC algorithms brings some challenges to the convergence analysis of the system.…”

Multi-lagged-input information enhancing quantized iterative learning control

“…An extended result for linear discrete-time systems has been proposed in Huo and Shen (2019b) by combining random data dropouts. In Zhang and Shen (2018), the quantized ILC problem using uniform quantizer with encoding and decoding method is investigated for linear and affine nonlinear systems.…”

Improving Boundary Level Calculation in Quantized Iterative Learning Control With Encoding and Decoding Mechanism

“…However, there is a major drawback in these studies: the logarithmic quantizer in the zero neighborhood requires infinite quantization precision, which is difficult to implement practically. To facilitate the engineering applications, a recent paper [26] uses a uniform quantizer that considers infinite and finite quantization level. Furthermore, to gradually improve the quantization precision, an encoding and decoding mechanism is introduced.…”

“…However, the problem in [26] is that when the quantization level is finite, there is an exponential term α N inside the quantization level calculation, which leads to the calculation results to be extremely large. In fact, the actual quantization level we need is fairly small, so this observation results in the following question: can we employ a rigorous estimation method to compress the calculated quantization level into a very small range so that it is sufficiently close to the actual maximum input of quantizer?…”