Day-Ahead Forecasting of the Percentage of Renewables Based on Time-Series Statistical Methods

Abstract

Forecasting renewable energy sources is of critical importance to several practical applications in the energy field. However, due to the inherent volatile nature of these energy sources, doing so remains challenging. Numerous time-series methods have been explored in literature, which consider only one specific type of renewables (e.g., solar or wind), and are suited to small-scale (micro-level) deployments. In this paper, the different types of renewable energy sources are reflected, which are distributed at a national level (macro-level). To generate accurate predictions, a methodology is proposed, which consists of two main phases. In the first phase, the most relevant variables having impact on the generation of the renewables are identified using correlation analysis. The second phase consists of (1) estimating model parameters, (2) optimising and reducing the number of generated models, and (3) selecting the best model for the method under study. To this end, the three most-relevant time-series auto-regression based methods of SARIMAX, SARIMA, and ARIMAX are considered. After deriving the best model for each method, then a comparison is carried out between them by taking into account different months of the year. The evaluation results illustrate that our forecasts have mean absolute error rates between 6.76 and 11.57%, while considering both inter- and intra-day scenarios. The best models are implemented in an open-source REN4Kast software platform.

The Effect of Using Aerosol Variables on the Performance of Deep Learning-based GHI Forecasting Models

“…Two research gaps are recognized in this section. The first is the lack of comparative studies on DL-based GHI forecasting models' performance with and without exogenous inputs, as what has been carried out in the literature with statistical models [12][13][14][15] and traditional ML models [16][17][18][19][20]. The second is the dearth of research on the effect of aerosol measurements, as one of the exogenous inputs, on ML-based GHI forecasting models' performance, as what has been done with physical and statistical GHI forecasting models [25][26][27][28][29][30].…”

Dynamic Pricing for Charging of EVs with Monte Carlo Tree Search

“…In the dataset, the start times and duration appear uncorrelated (Pearson's correlation coefficient = 0.008), so we assume these distributions to be independent. For more details about calculating the Pearson's correlation coefficient, the interested readers can refer [42]. Furthermore, the charging sessions do not go beyond 23:59; we also make this assumption in our problem instances.…”

Parachute strategy: hedging proposal for the Mexican stock market in the face the arrival of omicron

“…También el modelo SARIMA ha sido mejorado, con el modelo de media móvil integrado autorregresivo estacional con regresor externo (SARIMAX), al incorporar un parámetro exógeno en el modelo SARIMA, lo cual se ha comprobado empíricamente en investigaciones recientes, por ejemplo, Jain et al, (2021) pronostican el número de contagios diarios por COVID-19 en India y Prilistya et al, (2021) realizan el pronóstico para las llegadas de turistas a Indonesia en tiempos de COVID-19, Dutta y Roy (2021) predicen los niveles de contaminantes del aire en interiores, Cheng et al, (2021) predicen la ocupación por horas del departamento de urgencias de un gran centro médico de Estados Unidos, Banaś y Utnik (2021) predicen el precio de madera de pino en Polonia, Manigandan et al, (2021) pronostican la producción y el consumo de gas natural en Estados Unidos. También existen diversos estudios sobre electricidad, como lo son los trabajos de: Basmadjian et al, (2021), Kim et al, (2021), Pooniwala y Sutar (2021), Brusokas et al, (2021), y Abunofal et al, (2021. De este último se observa un pronóstico para los precios futuros de electricidad en Alemania para evitar pérdidas económicas y maximizar ganancias.…”