Overhead transmission line icing flashover trip characteristic analysis and risk model

Icing flashover trip is the key factor in stimulating vulnerability of power grids and leading to large-scale blackout. Ice flashover trip characteristic analysis and risk modeling were carried out based on grid level in the study. According to icing flashover characteristic analysis, the critical values of ice bridged and off-bridged of insulation system were identified. Furthermore, state division principle and risk rating method were proposed. State recognition was conducted by fuzzy theory. Considering the small sample and multi-input characteristics of the data, structural risk minimization principle was adopted, and a least squares support vector machines (LSSVM) risk model was built. Model parameters were optimized by Bayesian evidence reasoning. The comparative study of the proposed model with artificial neural network with error back propagation (BP-ANN) was proved that Bayesian-LSSVM had strong generalization ability. Finally, a set of vulnerability indices was used to analyze ice flashover trip vulnerability characteristics of the power grid.