Communication infrastructure effects on time detection of controlled islanding using robust neural network

Power system has been operated near its margins in restructured area. In such conditions, low frequency oscillations can make some islands corresponding to coherent groups. Many algorithms have been utilized to identify the time of islanding condition but in comparison, the neural network (NN) method shows more reliable and faster results. Recently, phasor measurement units (PMU) have presented new opportunities to analyzing power system by accurate and synchronized measures. These accurate data can be utilized to control the power system especially when dynamic state is detected. In this paper, a robust NN has presented by intelligent selecting PMU data. Impact of communication infrastructure failure, noisy measured data and delayed data are evaluated. IEEE 39-bus is used in DigSilent software to generate required data. Also neural network is simulated in MATLAB.