Predicting Chaotic Behavior in Electrical Circuits

Abstract

Chaotic behavior is a natural phenomenon that can be found all around us in our daily lives. This project is focused on analyzing the behavior in forced RL-Diode (resistor, inductor, and diode) electrical circuits. We determined that when the sinusoidal input voltage of the circuit was increased, the voltage across the diode experienced period doubling, quadrupling, and then eventually chaos. Furthermore, this project is focused on predicting when and how these chaotic properties emerged from data that we collected. The primary machine learning technique that is used to predict chaos properties is a recurring neural network called an echo state network. Accurately predicting chaos in a smallscale electrical circuit like our RL-Diode circuit and further research on this topic could lead to a greater understanding of chaos theory and its applications in machine learning and electrical engineering.