Neural sliding-mode load frequency controller design of power systems

Load frequency control (LFC) is one of the most profitable ancillary services of power systems. Governor dead band (GDB) nonlinearity is able to deteriorate the LFC performance. In this paper, controller design via a neural sliding-mode method is investigated for the LFC problem of power systems with GDB. Power systems are made up of areas. In each area, a sliding-mode LFC controller is designed by introducing an additional sate, and a RBF neural network is utilized to compensate the GDB nonlinearity of the area. Weight update formula of the RBF network is derived from Lyapunov direct method. By this scheme, not only the update formula is obtained, but also the control system possesses the asymptotic stability. Simulation results illustrate the feasibility and robustness of the presented approach for the LFC problems of single-area and multi-area power systems.