LINEAR OPTIMAL REGULATOR BASED ON FUZZY-LOGIC NEURAL NETWORKS

A Linear/Quadratic Optimal Regulator (LQR) carl adequately stabilize a nonlinear plant when the operating point is close to the one chosen f or linearization and design. A Fuzzy-Logic Neural Network (FLNN) has a learning ability to improve its performance over time by interaction with its environmental. This paper proposes to use a combination of LQ R and FLNN in order to enlarge the dynamic stability region of a nonli near system. Employing a synchronous-machine/infinite-bus as a nonline ar system, the effectiveness of an FLNN-based LQ regulator is evaluate d by comparing its performance to that of the classical LQ regulator. It is tested for small disturbance of the reference voltage V-ref and for a large disturbance caused by a large change in the load supplied by the machine.