Résumé:
Traditional power grids, with one-way communication, lack flexibility and efficient fault
management. On the other hand, Smart Grids integrate advanced technologies, allowing bidirectional
energy management and better adaptation to needs, while optimizing the integration of renewable
energies. However, their interconnection makes them vulnerable to cyberattacks, which can lead to
outages and risks for critical infrastructures. Securing these networks is therefore essential. This thesis
proposes an intrusion detection system based on machine learning, combining CNN and LSTM, to
improve the security of Smart Grids. Tested on the KDD99 dataset, the model showed good
performance in terms of accuracy, outperforming existing methods.
