Automatic extraction method suitable for deriving load model parameters

Abstract

Inappropriate load models could cause discrepancies between the measured and simulated responses in both the steady-state and transient state. Therefore, more accurate load models and their parameters need to be derived with the aid of measured data. Although more sophisticated measurement devices have been developed, the whole measured data such as for 30 seconds should not be used for deriving load model parameters, because the natural change in load structures regardless of voltage- and frequency-dependent load can deteriorate the accuracy of the derived voltage- and frequency-dependent load model parameters. In order to extract measured data that do not include the natural change in load structure, an automatic extraction method suitable for deriving load model parameters using a Fuzzy Inference System (FIS) is developed. The suitable data length can be specified using the correlation index between active power load and load bus voltage provided by the FIS. The measured data which are not used for the learning algorithm are used to validate the performance of the developed method.