Real-time data reassurance in electrical power systems based on artificial neural networks

Power system security is vulnerable to cyber attacks that may cause significant damages to the power grid and result in huge financial losses. In this paper, we show the risks associated with cyber attacks and propose an artificial neural network-based protection approach. The proposed algorithm can monitor the output of power flow calculations and detect data anomalies in real-time. The network observability rules are formulated as a mixed integer linear program (MILP) problem. The results of the MILP problem are used to decrease the amount of data input required by the algorithm while the system stays observable. We run our experiments on the IEEE 24-bus reliability test system. The experimental results show that the developed algorithm is a promising enhancement to ensure data integrity in control centers.

► We show the risks associated with a cyber-attack to a power system.
► We propose an artificial neural network-based protection approach.
► We use the network observability rules to reduce the size of the neural network.
► We run our experiments on the IEEE 24-bus reliability test system.
► Experimental results showed promising enhancement to ensure data integrity in control centers.