Adaptive Neural Network Command Filtered Backstepping Control for the Underactuated TORA System

An adaptive neural network command-filtered backstepping design algorithm is proposed for the underactuated translational oscillator with a rotating actuator (TORA). The system dynamics are transformed into nonlinear cascade through global change of coordinates. Considering weak sinusoid-type interaction and affine-free appearance in model, TORA looked at as pure feedback system. Two networks used to approximate unknown functions command filter produce virtual control its first second-order derivatives overcome explosion complexity problems backstepping. A error compensation dynamic designed influence on performance. Taking full account model structure TORA, each step subsystem reduce steps simplify process. stability proved Lyapunov theorem verified by numerical simulations.