Topology Identification in Low-Voltage Grids using Neural Networks

IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm), pp. 568–574

Abstract

As the number of decentralized generation plants and new customer types in the low-voltage level increase, distribution grid operators need to gradually move to active grid operation based on measurement data collected by smart metering systems and grid status analyses. However, in many cases, the necessary grid topology is not available to distribution grid operators in sufficient quality. This is necessary to determine active operating states on grid models, and thus to determine statements about the current degree of utilization of the grid. This paper presents a methodology for topology identification using a machine learning approach, which includes a Long Short-Term Memory model, which is a specialized type of Recurrent Neural Networks. Within the scope of hyperparameter sweeps, an optimal parameterisation of a trained model is identified. The trained model is then used to determine the topology of realistic low-voltage grids. A comparison of the results with an approach based on the correlation analysis of voltage time series shows no significant differences in prediction accuracy and is thus a promising initial step for future work.

Authors 4

1. Franziska Tischbein Aachen Digitalization and Energy Economics (IAEW)

   RWTH Aachen University

   Affiliation as printed

   Digitalization and Energy Economics (IAEW) at RWTH Aachen University,Institute of High Voltage Equipment and Grids,Aachen,Germany

2. Antoni Chajan Aachen Digitalization and Energy Economics (IAEW)

   RWTH Aachen University

   Affiliation as printed

   Digitalization and Energy Economics (IAEW) at RWTH Aachen University,Institute of High Voltage Equipment and Grids,Aachen,Germany

3. Luis Böttcher Aachen Digitalization and Energy Economics (IAEW)

   RWTH Aachen University

   Affiliation as printed

   Digitalization and Energy Economics (IAEW) at RWTH Aachen University,Institute of High Voltage Equipment and Grids,Aachen,Germany

4. Andreas Ulbig Aachen Digitalization and Energy Economics (IAEW)

   RWTH Aachen University

   Affiliation as printed

   Digitalization and Energy Economics (IAEW) at RWTH Aachen University,Institute of High Voltage Equipment and Grids,Aachen,Germany