LVQ Neural Network for Online Identification of Power System Network Branch Events

The transmission network is the backbone of the power system. Fast power system network branch event identification is required to maintain security of the power system. Phasor measurement units (PMUs) provide synchronized voltage and current phasor measurements to a centralized phasor data concentrator (PDC). Thus, PMU measurements can be used for online branch event identification. A fast and efficient approach is required to ease the computational load. Learning vector quantization (LVQ) neural network algorithm is presented in this paper as a solution to identify branch events. The power system is highly distributed. Thus, a cellular computational network (CCN) is utilized to distribute a single large LVQ's computational load. The IEEE 12-bus benchmark power system is used to illustrate the CCN-LVQ approach. The power system is simulated on a realtime digital simulator. CCN-LVQ approach is shown to be fast, efficient and comparatively accurate for transmission network branch event identification.