A cooperative CMAC neural network for hydro-generating system with doubly fed induction generators

This paper proposes an approach of cooperative cerebellar model articulation controller (CMAC) neural network that is based on the concept of combining CMAC and the adaptive linear neuron controller. The main purpose is to eliminate the excess self-learning and chattering phenomena of the traditional feedforward cooperative CMAC. In the proposed approach, the CMAC adopts the dynamic errors and desired output values of the controlled plants for the input vectors, it is parallel to the adaptive neuron controller which can play an important role in improving the robustness and adaptive property of CMAC. In order to verify the control quality of the proposed cooperative CMAC neural network, the digital simulation of the steady state regulation characteristics for the multivariable and nonlinear control of the hydro-generating system with doubly fed induction generators (DFIG) is studied. By comparison the traditional proportional-integral-differential (PID) controller, the simulation results have shown that the proposed scheme is of good adaptability and robustness under the generator parameter variation, the hydro-turbine model changed and external load disturbance.