Study on new neural network adaptive control of industrial manipulators based on desired compensation

A new Cerebellar Model Articulation Controller (CMAC) neural network (NN) adaptive controller is proposed for the tracking control of the uncertain Industrial Manipulators based on desired compensation, which approach the desired model smoothly by using expended 2th-order B-spline CMAC NN. The desired inputs eliminate the assumption on CMAC NN inputs which needed in conventional neuro-adaptive controller. The nonlinear feedback term is introduced to fully counteract the compensated unmatched terms, and a simple adaptive robust term is used to eliminate the unknown approach error and bounded disturb. Finally, the closed-loop global stability is guaranteed, and the validity of the proposed scheme is demonstrated through the simulation study on a two degree of freedom (DOF) Industrial Manipulator.