Project-based learning of Feedforward Neural Networks with a physical pendulum

In this paper the design and implementation of a Feedforward Neural Network-based controller on a pendulum are presented to demonstrate an example of a hands-on project that can be applied at the graduate and undergraduate levels in data-driven control courses. The project is planned in four main steps: system identification, PID control implementation, inverse model identification, and Feedforward Neural Network controller implementation. These steps are explained in detail in the paper, along with the models and control structures, and the mathematical expressions and algorithms are provided for use in physical deployment. The contribution of this paper is a detailed description of the design process of the controller using Neural Networks and a PID.