High-order ILC methods (Bien and Huh 1989; Chen et al. 1997; 1998; Hätönen et al. 2006; Gunnarsson and Norrl 2006) have attracted continuous attention for achieving a better control performance by using additional error information from previous cycles, although more memory storage is required for the high-order algorithms as a trade-off. The basic idea in Bien and Huh (1989), Chen et al. (1997), Chen et al. (1998), Hätönen et al. (2006), Gunnarsson and Norrl (2006) is to utilize more historical data collected from previous iterations to enhance the online control performance. However, it is noteworthy that online data in the current iteration, compared with the historical data in the previous iterations, can reflect varying characteristics of a process and the process disturbances in real time. Therefore, whenever possible, one should incorporate the historical data as well as the online data to achieve a better control performance. Furthermore, the existing higher order methods mainly focused on PID-type and optimization-based schemes, and most of their design and analysis are also limited to linear systems. To evaluate the performance of high-order ILC, some works (Bien and Huh 1989; Chen et al. 1997, 1998; Hätönen et al. 2006; Gunnarsson and Norrl 2006; Chi et al. 2015a, b) have been done by simulation study to illustrate that the high-order ILC can achieve a better performance with enhanced robustness than the low-order one. However, in theory, the question whether the control performance of high-order ILC is superior or inferior to that of the low-order ILC is still challenging. In Xu and Tan (2002a, 2003), it was shown that the low-order ILC was faster than the high-order ILC, but the discussion only focused on the linear learning controller for linear systems in the sense of time-weighted norm. Xu and Tan (2002a, 2002b, 2003) have also pointed out that based on the heuristic knowledge, it is possible to choose a high-order ILC scheme to achieve faster convergence for a certain norm. The reference Schmid (2007) has pointed out that up to now there is no theoretical proof about whether the high-order ILC works are better with faster convergence than the low-order one. Therefore, how to utilize more control information of previous iterations to guarantee that the high-order algorithm outperforms the low-order one is significant and challenging in the existing ILC theory.
Higher Order Data-Driven Iterative Learning Control
Intelligent Control & Learning Systems
Data-Driven Iterative Learning Control for Discrete-Time Systems ; Chapter : 9 ; 155-169
2022-11-16
15 pages
Article/Chapter (Book)
Electronic Resource
English
Constrained Data-Driven Iterative Learning Control
Springer Verlag | 2022
|Event-Triggered Data-Driven Iterative Learning Control
Springer Verlag | 2022
|ESO-based Data-Driven Iterative Learning Control
Springer Verlag | 2022
|Iterative learning data driven strategy for aircraft control system
Emerald Group Publishing | 2023
|Data-Driven Iterative Learning Control for Discrete-Time Systems
TIBKAT | 2022
|