A distribution system state estimation for an efficient integration of electric vehicle charging infrastructure into low-voltage grids

Konferenz: Internationaler ETG-Kongress 2019 - ETG-Fachtagung
08.05.2019 - 09.05.2019 in Esslingen am Neckar, Deutschland

Tagungsband: ETG-Fb. 158: Internationaler ETG-Kongress 2019

Seiten: 6Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Ritt, Bendic (Stromnetz Hamburg GmbH, Bramfelder Chaussee 130, 22177 Hamburg, Germany)
Scheffer, Volker; Ipach, Hanko; Becker, Christian (Hamburg University of Technology, Institute for Electrical Power and Energy Technolgy, Harburger Schloßstr. 20, 21079 Hamburg, Germany)

Due to political reasons, a growth in the usage of electric vehicles (EV) is expected for the coming years. The consequence is an increasing amount of EV charging operations, where the charging devices are mainly connected to the low-voltage level of the distribution grid. A high simultaneity factor of the charging processes together with inherent high and constant load demands can lead to a higher stress on grid assets and in extreme situations to violations of grid operation constraints. For a resolution of this problem, direct load management represents a favorable instrument in the view of distribution system operators. An efficient load management scheme, which acts on the loads only in critical grid situations, demands a grid state estimation of high accuracy. A deteriorating factor for the estimation quality in low-voltage grids is the very low-availability of real-time measurements. This work presents a methodology for implementing an accurate state estimation scheme for low-voltage grids, without the need for real-time measurements and solely based on historical smart meter data. For the realization of the objective, the methodology consists of the synthesis of two concepts. The first concept implements a state estimation algorithm for low-voltage grids using three-phase line models. In order to handle the low-availability of real-time measurement data, a concept for the generation of pseudo-measurements flanks the first concept. Hereby, neural networks trained with historical smart meter data are used to estimate the real-time load behavior of household loads. A series of exemplary simulations in a typical low-voltage grid from the city of Hamburg proves the applicability of the developed concept.