Detection of Cyber-Attacks and Power Disturbances in Smart Digital Substations using Continuous Wavelet Transform and Convolution Neural Networks

Smart Digital Substations use communication networks to exchange the information among its components to perform monitoring and control, which makes such components prone to cyber-security threats. The presence of power quality disturbances may add complexity to the problem of detecting such cyberattacks as some of these power quality disturbances behave in a similar manner to some attacks. In this paper, a novel approach that detects cyberattacks from power quality disturbances and normal operation is developed. The proposed approach uses only three out of the twenty-nine features to detect such cyberattacks. The proposed approach uses the continuous wavelet transform to represent such features in the time-frequency scalograms, which are then fed to a convolution neural network for detecting cyber-attacks and power disturbances in IEC-61850 substation systems. The proposed approach has been tested on three datasets from substation test systems as well as in realtime using OPAL-RT. The results have shown that the proposed approach was effective in detecting the cyberattacks from power disturbance and normal operation with a detection accuracy of 100% in simulation environment and 99.45% in real-time using OPAL-RT.