EVALUATING SEQUENCES OF EVENTS BASED ON NEURAL NETWORKS AT BULK SUBSTATIONS

This paper describes a neural network based intelligent system for rec ognizing and analyzing Sequence of Events (SOE) generated by substatio n Remote Terminal Units (RTUs). The main aim is to generate a short Di spatcher Quick Information (DQI) message and display it on the man-mac hine interface computer (MMI) with high priority to help the substatio n staff in understanding difficult situations. The system applies para llel Backpropagation Neural Networks (BPN) according to different prot ection zone types. The structure and a BPN of a special zone (120 kV t ransmission line) are described. Training method, coding input data an d testing the network are discussed in detail.