Browsing by Author "Galarza, Ainhoa"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item Ensemble forecaster based on the combination of time-frequency analysis and machine learning strategies for very short-term wind speed prediction(2023-01-01) Rodríguez, Fermín; Alonso-Pérez, Sandra; Sánchez-Guardamino, Ignacio; Galarza, Ainhoa; Tecnalia Research & Innovation; DIGITAL ENERGYBased on the predictions of fossil fuels depletion in the following years, as well as their negative impact due to generated exhaust fumes, eco-friendly generators, and more specifically wind generators, have arisen as a solution for the electric demand challenge. Wind energy consists in extracting energy from wind speed, and because of the uncertain and intermittent behaviour of this meteorological parameter, wind turbines output power cannot be optimally exploited. Although the vast majority of the research in wind speed forecasting field has consisted in the purpose of novel algorithms, these studies have not made a pre-processing step of the data in order to try to extract the maximum information from databases. Therefore, the goal of this paper consists in analysing whether the combination of time-frequency decomposition of wind speed data with different machine learning algorithms can increase the accuracy of current wind speed predictions for 10 min ahead. Obtained error metrics demonstrated that the deviation of developed wind speed forecaster was lower than 0.1% in 62% of the validation database. In addition, the root mean square error of the final forecaster was 0.34 m/s. This means an accuracy increase of 51.5% if the result is compared with benchmark model's results.Item Forecasting intra-hour solar photovoltaic energy by assembling wavelet based time-frequency analysis with deep learning neural networks(2022-05) Rodríguez, Fermín; Azcárate, Iñigo; Vadillo, Javier; Galarza, Ainhoa; Tecnalia Research & Innovation; DIGITAL ENERGYDue to the expected lack of fossil fuels in near future as well as climate change produced by greenhouse effect as consequence of environmental emissions, renewable energy generation, and specifically solar photovoltaic generation, has become relevant in present energy generation challenge. Photovoltaic generators have strong relationship with solar irradiation and outdoor temperature in energy generation process. These meteorological parameters are volatile and uncertain in nature so, unexpected changes on these parameters produce variations on solar photovoltaic generators’ output power. While many researchers have been focused in recent years on the development of novel models for forecasting involved meteorological parameters in photovoltaic generation, they commonly do not consider an analysis step of the data before using it in the developed models. Hence, the aim of this study consists in assembling a wavelet based time-frequency analysis of the used data with deep learning neural networks to forecast solar irradiation, in next 10 min, to compute solar photovoltaic generation. Results of the validation step showed that the deviation of the proposed forecaster was lower than 4% in 90.60% of studied sample days. Final forecaster's root mean square error was 35.77 W/m2, which was an accuracy improvement of 37.52% compared against persistence benchmark model.Item Very short-term parametric ambient temperature confidence interval forecasting to compute key control parameters for photovoltaic generators(2022-06) Rodríguez, Fermín; Insausti, Xabier; Etxezarreta, Gorka; Galarza, Ainhoa; Guerrero, Josep M.; Tecnalia Research & Innovation; DIGITAL ENERGYIn recent years, various forecasters have been developed to decrease the uncertainty related to the intermittent nature of photovoltaic generation. While the vast majority of these forecasters are usually just focused on deterministic or probabilistic prediction points, few studies have been carried out in relation to prediction intervals. In increasing the reliability of photovoltaic generators, being able to set a confidence level is as important as the forecaster's accuracy. For instance, changes in ambient temperature or solar irradiation produce variations in photovoltaic generators’ output power as well as in control parameters such as cell temperature and open voltage circuit. Therefore, the aim of this paper is to develop a new mathematical model to quantify the confidence interval of ambient temperature in the next 10 min. Several error metrics, such as the prediction interval coverage percentage, the Winkler score and the Skill score, are calculated for 95%, 90% and 85% confidence levels to analyse the reliability of the developed model. In all cases, the prediction interval coverage percentage is higher than the selected confidence interval, which means that the estimation model is valid for practical photovoltaic applications.