Multilayer artificial neural networks for real time power system state estimation

State estimation is a vital apparatus in observing the power electric grids. As the measure of the electric power grid keeps on growing, a state estimator must be all the more computationally effective and robust. This paper presents a real time state estimation using a new methodology of multilayer neural networks exhibited in composite topologies, hybrid Cascade and hybrid Parallel topologies in order to improve the estimation performance. The intent is to address the conduct of various composite topologies to contrast the robust performance indices by the maximum relative error, mean absolute percentage error (MAPE), root mean square error, and mean square error (MSE). The performance of distinctive topologies are contrasted with distinguish the best connection structural. The estimation performance of the proposed method is evaluated using real time data from the American Electric Power System in the Midwestern US which is published by the official website of University of Washington.