Deep Machine Learning Model-Based Cyber-Attacks Detection in Smart Power Systems

In this study, a deep learning-based attack detection model is proposed to address the problem of system disturbances in energy systems caused by natural events like storms and tornadoes or human-made events such as cyber-attacks. The proposed model is trained using the long time recorded data through accurate phasor measurement units (PMUs). The data is then sent to various machine learning methods based on the effective features extracted out using advanced principal component analysis (PCA) model. The performance of the proposed model is examined and compared with some other benchmarks using various indices such as confusion matrix. The results show that incorporating PCA as the feature selection model could effectively decrease feature redundancy and learning time while minimizing data information loss. Furthermore, the proposed model investigates the potential of deep learning-based and Decision Tree (DT) classifiers to detect cyber-attacks for improving the security and efficiency of modern intelligent energy grids. By utilizing the big data recorded by PMUs and identifying relevant properties or characteristics using PCA, the proposed deep model can effectively detect attacks or disturbances in the system, allowing operators to take appropriate action and prevent any further damage.