ROUGH DRAFT authorea.com/70540
Main Data History
Export
Show Index Toggle 30 comments
  •  Quick Edit
  • Comparison of deterministic and stochastic approaches for day-ahead solar power forecasting

    Abstract

    PV power forecast could mitigate the effects of high solar power injection into the Electricity Grid. Two main methods are currently used for PV power generation forecast. A deterministic approach that uses physical based models requiring detailed PV plant information. A data-driven approach based on statistical or stochastic machine learning techniques needing historical power measurements. The main goal of this work is to analyze the accuracy of these different approaches. Two original deterministic and a stochastic model for day ahead PV generation forecast were developed and a detailed error analysis was performed. Four years of site measurements were used to train and test the models. NWP data generated by the WRF models were used as input . Additionally a new parameter, called Clear Sky Performance index, is defined. It is the equivalent of the Clear Sky index for PV power generation forecast and it is here used to develop the stochastic model and an outperforming persistence model. The stochastic model not only was able to correct NWP bias errors, but also provides a better transposition. The deterministic approach leads to a skill score of 35% with respect to the persistence model, while the stochastic approach provides a skill score of 39%.

    Introduction

    In the last decade, cumulative installed capacity of photovoltaic (PV) has grown at an average rate of 49% per year, reaching by the end of 2014 a worldwide installed capacity of 177 GW. In 19 countries the annual PV contribution to electricity demand was estimated to exceed the 1% mark, with Italy leading with at least 7.9% followed by Greece at 7.6% and Germany at 7%. Different IEA scenarios predict for 2050 a PV penetration between 6% and 11% of the world electric consumption (IEA 2014, IEA 2014a).

    Electricity load can be affected by high PV generation, introducing a stochastic variability dependent on meteorological conditions (integration 2015). In particular, on the daily time scale, the PV production increases the rapidity of the load ramps so that a greater secondary reserve and ready supply is needed. This is accentuated in the evenings when the rapid reduction of large amounts of PV power is added to an increase in consumption (that brings to the night peak).

    Thus, the large share of PV power introduces new challenges for the stability of the electrical grid, both at the local and national level, requiring the need of more reserves to ensure electrical balancing and overcome the unpredictability and variability of the electricity demand. Moreover it implies an increase in costs related to transactions on the Day-Ahead and Intraday Energy Market and dispatching operations on the Real-Time Energy Market. Despite the challenges, grids can sustain high penetration of distributed power generation provided that quality of supply is addressed at connection point through the capabilities of modern power electronics, distributed control, and the use of ancillary services.

    PV power forecasts could mitigate the effects of high solar power injection into the electricity grid, both on grid management and on the energy market. Short-term forecasts (intra hours) could be use