Optimisation of the Use of Multiple Sources of Data in Short-term Photovoltaic Generation Forecasting Models

In a context of natural resources depletion, weather-dependent renewable energy sources play an increasingly important role in the electricity generation mix. Yet, high shares of renewables can jeopardise the safe operation of the power grid due to their variable nature. To address this challenge, it is essential to know the future amount of energy produced to balance production and consumption. In this thesis, we explore two main approaches that aim at improving the accuracy of short-term photovoltaic generation forecasting. The first option is to extend the existing statistical models found in the literature through the coupling with a physics-based model, and by operating a shift from static to weather-adaptive models. The second option lies in extending the range of available sources of information. In this regard, an in-depth quality analysis of production measurements emphasises relevant information, and exhibits the spatiotemporal correlations that may exist between the inputs. (author)