Neural-based control and stability of a nonlinear, non-autonomous, base-excited inverted pendulum system

Presents a novel application of multilayer neural networks for online control of a nonlinear, nonautonomous base-excited inverted pendulum system. The pendulum has two degrees of rotational freedom and its base-point moves freely in the 3D space. The goal is to apply control torques to keep the pendulum at the desired orientation. Four three-layered neural networks are trained online, using the backpropagation technique, to approximate the unknown nonlinear pendulum. The networks are employed within a controller to represent the inverse dynamics of the plant. The requirement of training accuracy, to guarantee the stability of the closed-loop system, is established using Lyapunov stability theory. The examination of the proposed controller, through simulations, demonstrates the promise of the methodology and exhibits positive aspects which cannot be achieved by previously-developed controllers