With currently widely available technology, we cannot store electricity in large quantities, therefore it requires consistency between consumption and generation of electricity for optimal operation of the electric power system. Until couple of years ago, the power system operated according to the established models of electricity generation in centralized power plants and distributing electricity to consumers. Nowadays, integration of distributed energy resources (DER), increasing prevalence of electromobility, expectation of more efficient use of energy and a more active role for consumers, who are prepared to change their habits of consumption, all bring new challenges for power system planners
The amount and dynamics of the electricity generated by DER in the network depend to a large extent on the weather factors. An increasing proportion of DER, especially renewable energy sources, does not only bring major challenges in the field of power system management, but it also affects the electricity market. Therefore it is very important to make forecasts of consumption and generation of electric energy, especially for DER, since it’s crucial for the stable operation of power system and for adaptation of generation of conventional sources to consumption and also for trading with electrical energy.
In the master’s thesis, the subject of interest is the analysis and short-term forecast of (active) power flow through substation Breg. It is characteristic for the network of substation’s Breg, that it has installed a large number of DER. Their influence, especially solar power plants, which have almost
87 % of total power of DER, is the most noticeable on the daily diagram of power flow. Elektro Maribor d.d. gave us historical data of the power flow through substation Breg and the generation of solar power plants in that network.
We tackle the forecasts with the use of ARIMA (autoregressive Integrated Moving Average) model and its combinations respectively. In the master’s thesis we describe time series and their components, followed by a description of key concepts, which are needed in the analysis of time series. Further it describes the methodology of the ARIMA model, which belongs to the group of statistical methods. The building and selection process of the most appropriate ARIMA model for short-term forecasting for one day ahead is explained. At the end, a comparison of all three methods of forecasting power flow is given. The first method is based on a combination of forecasts of production and generation of electrical energy, the second method is based only on historical data of power flow, the third method uses exogenous time series of global irradiance in addition to historical data of the power flow.