Rotary inverted pendulum (RIP) is a well-known system that is commonly employed as an ideal benchmarking model for verifying linear and nonlinear control algorithms thanks to unique unstable and highly nonlinear natures. In this paper, an intelligent control method is developed for stabilizing such the RIP system in upright posture. The controller is structured from a linear quadratic regulator (LQR) and an online radial basis function (RBF) Neural-Network compensator. The LQR term plays a crucial role in yielding the nominal control signal based on a linearized model. Meanwhile, the neural control term is adopted to suppress the systematic deviation and external disturbances as the system is far from the equilibrium state. A damping segment-wise adaptation rule is proposed to activate the network operation. Stability of the closed-loop system is then proven by Lyapunov analyses. Effectiveness and feasibility of the advanced controller are confirmed throughout comparative simulation and real-time experiments.

No abstract

No abstract