A Practical Approach to Construct a Digital Twin of a Power Grid using Harmonic Spectra

The electrical energy system is transforming itself into a sustainable energy supply in response to the decline in fossil fuels. This conversion is driving the expansion of renewable energy facilities such as photovoltaic plants and wind power plants. Except for large hydropower plants and offshore wind farms, the integration of renewable energies takes place predominantly in the medium- and low-voltage distribution networks. This leads to a lack of observability and increasing grid complexity. Consequently, distribution network operators are constantly faced with a challenge in terms of observability. A comprehensive installation of measuring instruments in the medium- and low-voltage networks has proved economically unviable. An alternative approach to network state monitoring within the framework of power grid digital twin (DT) is therefore developed in this paper. The patterns of the electrical energy system are detected and modeled employing an artificial neural network (ANN) in connection with the associated harmonic spectra. Based on this DT model, the active powers of renewable energy facilities are estimated through the measured voltage data. In this regard, this work is first devoted to the modeling of an ANN-based DT estimator. The proposed power state estimation is then validated with the measured data from a field test. The accuracy of the estimation will be investigated according to the different influencing factors.