| Abstract [eng] |
First electricity market liberalization processes in Europe had started in 1990. Since then market had changed significantly. Nowadays, electricity market is going through the modernization period of its lifetime. Smart-grid and smart-metering systems are developing and becoming more available in the market. Smart meters are extremely important for establishing permanent remote connection between electricity suppliers and consumers. As a result, electricity suppliers are provided with a huge amount of data that is highly granular. Using this data effective demand response programs could be introduced in the market. It is proven that demand response programs can have an effect on impact on the final electricity demand. In that case, it must be considered when the final electricity demand forecast is done. As demand response programs are not triggered frequently point forecasts could be rough, in that case – probabilistic output used in this paper. In order to evaluate the impact of demand response programs to the actual loads a few steps taken. First, historical load based on weather data compared to the actual load, affected by demand response program implemented. By comparing these datasets, we can obtain an indication of how much electricity demand reduced the final load. Second, forecasting model for historical load based on weather data created. Third, forecasting model created for what we have learned from the first step. Finally, the second and the third steps combined. Forecasting models are based on artificial neural networks that were implemented by using MATLAB software with the Neural Network Tool. The national electricity distribution company provides data about the average consumption rates recorded during the trial project of smart-metering implementation. Lithuanian hydro meteorological service company provides historical weather data. Results shows interval probabilistic forecasts provide more informative and meaningful results than point forecasts. Interval forecasts could be used for re-evaluating current electricity pricing models, setting-up demand response programs, various risk management. In addition, calculated that during the trial project of smart-metering implementation electricity demand reduced by 11%. |
