Pseudo-worst-case forecast for a preventive control in LV smart grids

Conference: NEIS 2020 - Conference on Sustainable Energy Supply and Energy Storage Systems
09/14/2020 - 09/15/2020 at Hamburg, Deutschland

Proceedings: NEIS 2020

Pages: 6Language: englishTyp: PDF

Balouchi, Razieh; Weisenstein, Marco; Wellssow, Wolfram H. (University of Kaiserslautern, Kaiserslautern, Germany)

To increase the capabilities of monitoring and controlling of low voltage smart grids, input data availability has a significant role. Fortunately, the upcoming advanced metering systems (smart meter) infrastructure at the distribution grid level increases the amount of available input data significantly. However, this infrastructure is not suitable for real-time applications in general. Usually, the smart meters only send snapshot data in relatively long sample rates. Controlling a system only based on such snapshots would lead to problems. Forecasting the next step data would improve the control performance. However, the current and voltage profiles in the low voltage grid show high fluctuations due to the stochastic prosumer behavior, therefore, precise forecasting is not possible. Forecasting a more or less worst-case is more practical and can be used for preventive control to avoid possible improper grid states like overloads and over- or under-voltages. This paper presents a heuristic method to generate these pseudo-worst-case forecasts using historical data while considering the ongoing dynamic behavior of the observed data. The validation is done by simulation using the IEEE European LV test feeder model and synthetic one-year load profiles. In addition, a 193-node real low voltage grid located in Germany and real measured one-year load profiles have been analyzed. The results are promising. In at least 96 % of the simulated cases the values are within the range of the forecast. The average success rate is 98.16 %. Only in a small acceptable number of situations with extremely high gradients, the real value exceeds the forecast.