Wind power plants hybridised with solar power: A generation forecast perspective

The association of different variable renewable technologies in hybrid power plants and the benefits of their aggregation for the operation of power systems is an area of recent research. Accurate forecasts are crucial for extracting those benefits and promote an optimal integration of such plants into power systems and electricity markets. This study focuses on the hybridisation of existing wind power plants with different shares of solar photovoltaic capacity and investigates how these power plants can reduce their combined forecast errors and thus, increasing profitability in electricity markets. The work uses a forecast methodology based on a sequential forward feature selection algorithm which employs two different objective functions and an artificial neural network approach previously presented but, in this case, it is applied to the specific case of hybrid power plants. The methodology uses as input data from a numerical weather prediction model and iteratively selects meteorological features to achieve the different objective functions implemented, namely i) minimisation of the root mean square error; or ii) maximisation of the market remuneration. The methodology developed was applied to three case studies in Portugal with different levels of wind and solar generation complementarity. The results show that the hybrid power plants can increase market value by up to 5% and total remuneration can increase by up to 30% when compared with the existing wind power plant, while it is possible to reduce the forecast errors by nearly 4%. The obtained results highlight the need to select the most relevant meteorological features to maximise the accuracy of the power forecast and the renewable power producers revenues in a market environment.