Adaptive control of power systems using radial basis function network identification and predictive control calculations

An adaptive power system stabilizer using an on-line trained radial basis function (RBF) network and a pole-shift predictive controller is developed in this paper. The RBF-identifier is used to identify the system parameters in an on-line mode. In the proposed control, the multi-step ahead predictions are included in the overall performance index. The drawback of multi-step ahead optimization is the computational burden attached to it. In this paper simplifications are proposed using “dynamic control limits” wherein the control limits are not fixed at their absolute physical limits but are calculated on-line. These new control limits are passed to a pole-shifting numerical optimization routine to calculate a suitable control signal. The RBF-identifier and pole-shift coupled model is tested on a single-machine infinite bus power system model to verify its effectiveness