Cybersecurity Resilient for Quartile-based Transformer Differential Protection Scheme

This paper proposes a new cybersecurity resilient technique utilizing a two-layer feedforward neural network (2LFFNN). It is also able to predict the correct value of the altered SV packets. The performance of the proposed technique is evaluated in accordance with the quartile-based differential protection (87Q) scheme, which provides superior performance compared to the conventional transformer differential protection (87T) scheme. Nevertheless, as per the statistical analysis, the 87Q scheme is more vulnerable to cyber security threats in IEC 61850 enabled substation. The attacker can easily attack the IEC 61850 process layer and gain access to the SV messages, which carry current and voltage information of the substation equipment to the process layer. To test the proposed cybersecurity resilient scheme, 51000 cases of synthetic data, used for training 2LFFNN, pertaining to different types of attack are generated by modelling the 87Q scheme in the PSCAD/EMTDC software package. After training the neural network, the performance of the proposed 2LFFNN is verified during the FDI attack and SV packet loss/delay. The simulation results indicate that the proposed scheme along with 2LFFNN is not only capable of identifying cyber-attacks but also able to mitigate them.