Enhancing Cybersecurity in Smart Grid Systems Through Advanced Log File Analysis with Machine and Deep Learning Techniques

This research investigates the application of machine and deep learning techniques for enhancing cybersecurity in smart grid systems through log file analysis within digital forensics and incident response frameworks. Employing the NUSW-NB15_GT dataset, the study evaluates several machine learning models, including Random Forest, Gradient Boosting, and XGBoost, against deep learning architectures such as CNNs, LSTMs, and Transformers. The findings illustrate that machine learning models significantly outperform deep learning models, with XGBoost leading at an accuracy of 0.8933. The LSTM model shows moderate success with an accuracy of 0.8349. In contrast, the Transformer model lags with an accuracy of 0.5146, indicating its challenges with the structured, high-dimensional data typical of this dataset. This disparity highlights the superior capability of traditional machine learning methods in managing complex data interactions effectively, suggesting their preferential utility for smart grid cybersecurity. The results of this investigation establish a critical foundation for future advancements in protecting smart energy infrastructures and highlight the necessity for continued research into deep learning's potential in broader, more intricate datasets.