Improved representation of power system load dynamics using heuristic models

The importance of having accurate load models in power system stability studies has been well established in the literature as being essential for precise power system transient event investigations. The power industry presently uses composite load models in typical stability programs (e.g. LOADSYN and PSS/E). However, the parameters of the composite load models need to be tuned for each type of disturbance based on an assumed load composition, and are often inadequate for matching the modeled dynamics of the power system disturbance event to actual measured results. A stochastic time series technique in the form of an ARMAX mathematical model is presented in this paper as a novel alternative for dynamic load modeling. The model parameters are estimated using on-line measurement data for a number of disturbance events, collected from five substations in the Victorian electricity network in Australia. The performance of the model is then evaluated for other transient events, and compared against the recorded response for these events. The results achieved show that this heuristic-based model is robust and effective in predicting the dynamic response of a power system load across a range of events spanning various seasons and locations.