Design and implementation of adaptive neuro-fuzzy inference system for the control of an uncertain ball and beam apparatus

Abstract

Controlling an uncertain mechatronic system is challenging and crucial for its automation. In this regard, several control-strategies are developed to handle such systems. However, these control-strategies are complex to design, and require in-depth knowledge of the system and its dynamics. In this study, we are testing the performance of a rather simple control-strategy (Adaptive Neuro- Fuzzy Inference System) using an uncertain Ball and Beam System. The custom- designed apparatus utilizes image processing technique to acquire the position of the ball on the beam. Then, desired position is achieved by controlling the beam angle using Adaptive Neuro-Fuzzy and PID control. We are showing that adaptive neuro-fuzzy control can effectively handle the system uncertainties, which traditional controllers (i.e., PID) cannot handle.