Abstract:
Power system protection and emergency control systems require continuous monitoring to prevent grid-wide frequency instability or system islanding. Amongst different meth...Show MoreMetadata
Abstract:
Power system protection and emergency control systems require continuous monitoring to prevent grid-wide frequency instability or system islanding. Amongst different methods, cellular computation networks (CCNs) are more suitable for providing dynamic frequency predictions. CCN is a scalable and distributed framework using the concept of cells and intercellular connections to represent a power system topology with multiple synchronous machines, transmission lines and loads. However, from a practical viewpoint, fast computing algorithms are needed for effective monitoring and response. In this paper, a lite CCN is proposed based on generalized neurons (GNs), referred to as the cellular generalized neuron network (CGNN). The performance of CGNN is compared with that of a cellular multilayer perceptron network (CMLPN). The frequency predictions by the CGNN and CMLPN have been validated on a two-area four-machine benchmark power system. Data for the CGNN is provided from a real time digital simulation of the power system with phasor measurement units (PMUs). Based on a performance metric, results obtained show that the CGNN outperforms the CMLPN, especially in terms of least number of trainable parameters, low training time and better prediction accuracy.
Date of Conference: 24-29 July 2016
Date Added to IEEE Xplore: 03 November 2016
ISBN Information:
Electronic ISSN: 2161-4407