Cellular Computational Networks based Voltage Contingency Ranking Regarding Power System Security

摘要

Contingency ranking in simple terms is a process deployed to select the most serious contingencies that happen to the power system in the event of unplanned component outages or topological changes. In this study, Cellular Jordan Recurrent Networks (CJRNs) based non-linear voltage prediction is used for contingency ranking. The CJRNs based predicted voltages are used in computing performance indices that give proximity to voltage collapse severity of contingencies. The problem is solved combining with concentric relaxation method to reduce computation time. In order to show the advantages of the proposed CJRNs method, single line contingency ranking is tested by CJRNs voltage prediction model trained by different time steps. A novel voltage prediction correction mechanism which is applied after the prediction process shows to be effective for all time steps based contingency ranking.