Detection of cyber-attacks on smart grids using improved VGG19 deep neural network architecture and Aquila optimizer algorithm

This study introduces an innovative smart grid (SG) intrusion detection system, integrating Game Theory, swarm intelligence, and deep learning (DL) to protect against complex cyber-attacks. This method balances training samples by employing conditional DL using Game Theory and CGAN. The Aquila optimizer (AO) algorithm selects features, mapping them onto the dataset and converting them into RGB color images for training a VGG19 neural network. AO optimizes meta-parameters, enhancing VGG19 accuracy. Testing on the NSL-KDD dataset generates remarkable results: 99.82% accuracy, 99.69% sensitivity, and 99.76% precision in detecting attacks. Notably, the CGAN technique significantly improves performance over GAN. Importantly, this method surpasses various deep learning techniques such as VGG19, CNN-GRU, CNN-GRU-FL, LSTM, and CNN in accuracy. Addressing the critical need for robust SG intrusion detection, our work merges Game Theory, swarm intelligence, and deep learning, yielding superior security accuracy. The novelty of this study is implanted in the integrated approach, distinguishing it from previous research and contributing to effective protection against cyber threats in smart grids.