Intelligent Hermite Neural Control for Reaction Wheel Inverted Pendulums

Abstract

In order to solve the problem that the reaction wheel inverted pendulum (RWIP) is highly nonlinear, unstable and underactuated nature, this study proposes an intelligent Hermite neural control (IHNC) system which has two main components: one is a speed controller and the other is a neural controller. The speed controller is designed to overcome the effect of reaction wheel rotation speed on the inverted pendulum. The neural controller, which employs a Hermite broad-learning neural network (HBNN), is designed to drive the pendulum at an upward unstable balance point and keep it controlled there. The HBNN extends the network structure in terms of width and internal feedback loop to achieve fast learning and dynamic mapping capabilities. Finally, the experimental results demonstrate the effectiveness of the proposed IHNC system.