Data‐Driven Model Free Adaptive Iterative Learning Control for Nonlinear Systems With Time Delay and Its Application to Multivariable Processes

ABSTRACT For multivariable systems with multiple time delays, significant research has been conducted on their stability; however, there are comparatively few studies on their speed. In this research paper, a dual‐input and dual‐output data‐driven model‐free adaptive iterative learning control (DD‐MFAILC) based on time delay is proposed for multi‐input, multi‐output (MIMO) nonlinear processes using input/output (I/O) data. The proposed DD‐MFAILC algorithm, originally designed for single‐input and single‐output (SISO) systems, has been extended to accommodate dual‐input and dual‐output control (DIDO) systems. Complete convergence and stability proofs have been provided, and its performance has been fully evaluated. A simulation test utilizing the DD‐MFAILC algorithm was performed on a typical numerical example, specifically focusing on the Wood‐Berry distillation column as a case study. Step signals were used for comprehensive performance testing. The proposed DD‐MFAILC algorithm demonstrated high tracking performance on the Wood‐Berry distillation column, achieving average tracking accuracies of 97.58% for the overhead product and 95.28% for the bottom product.