Gestion de la génération d’un système solaire autonome connecté à un micro-réseau électrique pour absorber les charges de pointe

Abstract

In the urgent context of the climate crisis, optimizing energy management has become an imperative to reduce our energy consumption and the carbon footprint associated with the use of non-renewable energies. The management strategies put in place by grid operators, such as the imposition of subscribed power, aim to encourage consumers to moderate their consumption to ensure the stability of the power network, while reducing their costs. However, for consumers, adopting efficient energy management often means changing long-established habits, and sometimes adopting complex and inefficient solutions. A common method of reducing electricity consumption in buildings is peak shaving, which compensates for part of the load, in particular to meet the subscribed power imposed by grid operators. This technique involves storing energy during off-peak periods and releasing it during peak periods. From a carbon footprint reduction point of view, the integration of battery energy storage systems with renewable sources, in particular solar energy, is proving to be an effective solution for lightening the load on the power grid while improving energy efficiency. However, due to the intermittency of these energy sources, effective management remains necessary. The main objective of this thesis is therefore to determine a predictive model of solar production capable of efficiently managing a photovoltaic system connected to a microgrid. Different modeling approaches have been explored, including photovoltaic modeling, regression methods, decision trees and combined techniques. The performance of predictive models was evaluated using measures such as MAE, RMSE and coefficient of determination (R2). The validated model was then successfully integrated into a microgrid testbed, with a detailed discussion of the technical, software and hardware aspects of this integration. In conclusion, this work has validated a robust methodology for predicting solar production and highlighted the importance of future research into weather forecasting models and historical data collection to improve the accuracy of energy forecasts. Access to extensive weather archives is seen as a key step towards fully exploiting the capabilities of solar power plant, making this work an important contribution to renewable energy management research.

Date	29 Apr 2024
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Louis-A. Dessaint (Supervisor)