Iteratively reweighted least squares for maximum likelihood identification of synchronous machine parameters from on-line tests

This paper presents a new approach for the statistical identification of synchronous-machine parameters from on-line test data that were recorded on a 202 MVA hydro-generator at Hydro-Quebec's La Grande 3 generating station. Data processing is performed to remove harmonics in noise-corrupted measurements. The time-domain parameter identification is carried out by means of our proposed maximum-likelihood estimation method, also called the iteratively reweighted least-squares algorithm. A comparison of the results with the ordinary weighted least-squares estimation, which is equivalent to the maximum-likelihood estimation only when the noise is white, shows the superiority of the proposed method. This procedure appears more convenient than previous schemes for parameter identification of the synchronous-machine linear equivalent-circuits, especially when the noise statistics are poorly known.