Invented pendulum stabilization using artificial neural network

Abstract

The article raised the problem of stabilization of electric vehicles. The inefficiency of control systems built on the basis of classical methods is considered. Mathematical modeling of the inverse pendulum, which was chosen as the control object, was carried out. Simulation modeling of the stabilization of the inverse pendulum in the "OpenAI gym" environment has been carried out. A comparative analysis of the effectiveness of using a control system based on an artificial neural network and a classical control system for the problem of stabilizing an inverse pendulum is carried out. The system of total rewards in the simulation environment was chosen as an efficiency criterion. An assessment of the effectiveness of using the virtual environment "OpenAI gym" for comparing classical algorithms and algorithms based on neural networks is given. A model has been developed in ROS (Robot Operating System) to further study the effectiveness of inverse pendulum stabilization algorithms.