Estimation of power system parameters

This paper describes the use of a system identification technique to estimate the parameters of a low-order, dynamic model for a power system. The basic technique is to obtain an ARIMAX model by applying a prediction error method to data consisting of 1-s samples of the system frequency and the power output of the Dinorwig fast-response pumped-storage station. From this model the natural frequency, damping factor and stiffness (or "beta") of the power system are obtained. The paper first establishes an appropriate order for the ARIMAX model. Simulation is used to validate the model and ensure that the results are not an artefact of either the input, the identification method or the model structure. It is shown that the precision of the parameter estimates is limited by unknown load disturbances. An example of the use of the technique to analyze the daily variation of parameters is given.