Fast procedure for constructing an accurate dynamic average-value model of synchronous machine-rectifier systems

The analytical derivation of the average-value models (AVMs) for synchronous-machine-fed rectifier systems depends on many factors and is challenging in general. Various models presented in the literature rely on a single operating mode and a dc filter configuration. A recently proposed parametric approach utilizes a detailed simulation as a basis to establish an AVM that uses a proper full-order qd generator model. This paper presents a procedure to extract the appropriate nonlinear algebraic functions to represent the rectifier averaged behavior. In contrast to existing methods, the proposed procedure does not require significant analytical derivations and/or computational resources and results in a highly accurate dynamic AVM. The proposed methodology is validated with the results of detailed simulation and hardware experiment, and is shown to be very accurate in predicting the large-signal time-domain transients as well as the small-signal frequency-domain characteristics.