Designing an adaptive neural network controller for TORA system by using Feedback Error Learning

As a special kind of nonlinear systems, underactuated systems are of great interest in both theoretical research and real applications. The TORA (Translational Oscillator with Rotational Actuator) is an underactuated system. This system was developed as a simplified model of a dual-spin spacecraft and rotary machines for investigating the resonance capture phenomenon. This paper presents a new method for control of TORA system by using Feedback Error Learning (FEL) scheme. Adaptive neural network was used in this scheme, and backpropagation techniques utilized as learning algorithm for stabilization of TORA. Simulation results are presented to show that the proposed FEL is able to stabilize the TORA system even in the presence of disturbance.