Challenges in deriving equivalent models of active distribution grids from operational data

Wide area monitoring systems equipped with phasor measurement unit technology allow measurement and logging of time-synchronous operational data inside modern power systems. The usage of those operational data also for deriving equivalent models of active distribution networks (ADN) for power system stability studies therefore comes naturally. This paper consists of the analysis of the behaviour of ADN and the analysis of the properties of common equivalent model types and procedures with regard to their parameterization using operational measurement data. In the analysis of the behaviour the quasi-stationary nonlinearity between the input and output variables of the active distribution network is proposed as a measure of its complexity or the complexity of the identification problem of a distribution network in general. The discussion of the emergence of this nonlinearity, depending on the behaviour of distributed generation (DG) controls, thus allows to generalize the findings collected here to other distribution networks of different composition. In the determination of equivalent models the properties and conditions for the parameterization of the quasi-stationary nonlinearity is discussed into the most common equivalent model types in literature, the black-box models based on artificial neural networks (ANN) and the grey-box models based on parametric models.