Blackout estimation by neural network

Abstract

Cascading failures play an important role in creation of blackout. These events consist of lines and generators outages. Online values of voltage, current, angle, and frequency are changing during the cascading events. The percent of blackout can be estimated during the disturbance by neural network. Proper indices must be defined for this purpose. These indices can be computed by online measurement from WAMs. In this paper, voltage, load, lines, and generators indices are defined for estimating the percent of blackout during the disturbance. These indices are used as the inputs of neural networks. A new combinational structure of neural network is used for this purpose. Proposed method is implemented on 39-bus New-England test system.