Detecting electromechanical oscillations in power systems

The detection of electromechanical oscillations in power systems and the estimation of their parameters (frequency f and damping xi ) is based on a linear time-varying model. The parameters of the linear model are on-line adapted by means of Kalman filtering techniques to approximate the measured signal y (representing the behaviour of the power system) optimally in the sense of a quadratic criterion. The estimated model parameters are then the basis for the calculation of parameters of the oscillations. Adaptive algorithms are based on a recursive calculation of the estimated parameter vector for each time-step based on the new value of the measured signal and the old values of the estimated parameters. As opposed to the collection of data over a time window and then performing the parameter identification at once, any change in the system can thus be detected much faster.