MIMO feedback linearization control for power systems

Effectiveness of a multi-input, multi-output (MIMO) feedback linearization controller (FBLC) for power oscillation damping is illustrated in this paper. Oscillatory behavior of the system is estimated online from the measured quantities using a special form of neural network compatible with the feedback linearization framework. Levenberg–Marquardt (LM) algorithm is adapted to operate in a sliding window batch mode for estimation of the neural network parameters. The coefficient vector in the FBLC formulation is updated adaptively using the projection algorithm to suit changing operation scenarios. A case study is presented on a reasonably large-scale power system having three critical oscillatory modes. Two power electronic actuators located on separate transmission lines are used to control these modes resulting in a MIMO controller. Proposed FBLC is shown to yield acceptable closed-loop dynamic response with very little information about the plant model. Performance of the FBLC is benchmarked against a conventional model based controller.