Fuzzy and optimal control of a two-link flexible manipulator

A linear quadratic Gaussian (LQG) strategy controls a two-link flexible robot manipulator tracking a two-dimensional square trajectory 12.6 m×12.6 m. Slew angles together with a Gaussian white process and white or non-white measurement noise are fed into a LQG regulator (Kalman filter). A FLC strategy incorporates two fuzzy controllers substituted for the LQR state-space dynamics equations. Trajectories were obtained for the LQG strategy with Gaussian white and non-white measurement noise. The trajectory obtained for white measurement noise closely approaches a perfect square while those obtained for non-white measurement noise deviate. The trajectory obtained with the FLC strategy is similar to that for LQG with white measurement noise. Fuzzy control is found to provide robustness in operation and can be constructed with less mathematical complexity than LQG. The deviation in trajectories for LQG with non-white measurement noise suggests sub-optimal control and possible instability. This study has demonstrated the extent of deviation in tracking with an LQG strategy for non-white measurement noise and the FLC strategy as a viable option for precise and robust tracking control of a two-link flexible robot manipulator