Parameter Estimation for Power Transformer Models from Time-Domain Data

The existence of power system harmonics has been realized for many years. With the advent of new technologies and more solid state loads, power system harmonics are certain to increase both in amplitude and frequency bandwidth. Signal propagation in power systems depends on transformer impedances. Transformer impedances, on the other hand, are functions of signal frequency as well as input voltage and loading condition. Present power transformer models with single-valued parameters obtained from tests cannot adequately represent the transformer in the presence of harmonics. Accurate modeling of transformers must account for changes in signal frequencies and loading conditions as they occur. This paper presents algorithms for estimation of the parameters of a structurally known transformer model. These algorithms use time-domain data and continuously update the estimates, thereby accounting for changes in operating conditions. The techniques presented here can be applied to three-phase and single-phase N-winding transformers. The results of model identification for a single-phase two-winding transformer is given. The data acquisition system used and the problem of identification in a noisy environment are discussed. The simulated response of the identified model is compared with the transformer actual response.