Nonsingular Terminal Sliding Mode Control for Active Power Filter Using Recurrent Neural Network

In this paper, a nonsingular terminal sliding mode control (NTSMC) scheme with a recurrent neural network (RNN) structure is proposed for an active power filter (APF). The RNN proposed in this paper can approximate unknown part of system more accurately than the traditional neural network (NN), owing to the feedback loops in its hidden layer, where weights and output signals are stored and regarded as feedback signals. Compared with the traditional NN with fixed centers and widths, the centers and widths of the RNN are updated in real time by adaptive laws, which are derived based on the Lyapunov stability theory and Taylor series, and eventually stabilize at the optimal value. In addition, a nonsingular terminal sliding mode controller is developed to ensure that sliding surface and equilibrium point can be reached in a short finite time and simultaneously avoid singular problems existing in ordinary sliding mode controller. Simulation studies are conducted to verify the effectiveness of the developed scheme.