Designing a Neural Network-Based Controller: A Practical Educational Tool for Data-Driven Control Courses

In this article, the design of a controller based on Artificial Neural Networks (ANN) as a practical educational tool for data-driven control is proposed. In data-based control courses, the project plays a crucial role in understanding control methods applied to non-linear systems. A controller based on ANN models the system dynamics to control and compensate for system uncertainties. Real data from a simple physical pendulum is used to design a NARMA-L2 controller for angular position trajectory tracking and uncertainty compensation. The results show that trajectory tracking control is achieved, allowing students to learn and understand the structure and application of data-based controllers. Potential future work may encompass refining the didactic experiment to optimize its effectiveness as an educational tool.