Auto-Disturbance-Rejection Controller Design Based on RBF Neural Networks

Abstract

Compared with general controllers, the auto-disturbance-rejection controller (ADRC) has been paid much more attentions in the field of control as it has advantages on simple algorithm, strong robustness, and less dependence on the mathematical models of the plants. However, the tuning procedure of the ADRC parameters is very complicated due to its reliance on the human experiences. Therefore, a radial basis function neural network (RBFNN) is introduced to regularize the ADRC parameters automatically on line. Simulation results indicate that the algorithm above is feasible.