Data-induced learning control for unknown nonlinear systems with full-tracking performances

In tackling the challenge of achieving high control accuracy while refining control effort in unknown nonlinear systems, this paper presents a novel data-induced learning control (DiLC) method. The DiLC method marks the inaugural use of operational data to iteratively update controller parameters for unknown nonlinear systems, thereby achieving full-tracking performances over the entire operation time interval. The DiLC method enriches the traditional iterative learning control (ILC) paradigm by integrating a feedback control-based approach. This integration enables exploratory trials during the initial iteration process and facilitates the online collection and iterative utilization of operational data from actual operations, thus gradually enhancing control accuracy. Furthermore, it offers a significant advantage over prevalent finite/fixed-time and prescribed performance methods in repeatable tasks. Inspired by leveraging operational data, it effectively mitigates the constraints imposed on the system and adjustment parameters by existing methods, hence providing a more flexible and efficient solution. The efficiency of the DiLC method in accuracy evolution is proven through error reduction profiles, confirming its prospect for full-tracking performances in unknown nonlinear systems.