Study on sustainable development of power transmission system under ice disaster based on a new security early warning model

Abstract As the ecological climate deteriorates, common disasters including typhoons, earthquakes and ice disasters occur frequently in southern China. In particular, ice disasters can make severe ice coating on power line, resulting in power outage accidents. Not only do power supply disruptions cause major economic losses, but also they threaten people's lives in extremely cold conditions. Therefore, studying on the security early warning of ice coating damage to power line is significant and urgent. This paper adopts grey correlation analysis (GRA) to obtain 16 influence factors, of which the correlation degrees are over 0.74. In order to reduce the internal relevance among the impact factors, 5 common factors are extracted as the predictive model input values through factor analysis. With the input weight and hidden layer threshold optimized, a new extreme learning machine based on the adaptive whale optimization algorithm improved by chaotic sine cosine operator (CSCWOA-ELM) is established to forecast the ice coating damage to power line in southern China. To verify the accuracy and advancement of the proposed model, real data from the power repair projects under ice disaster are selected for experiments. The simulation results prove that the extraction of common factors can highly improve the prediction accuracy by approximately 27.80%. Compared with the 3 benchmark models, the CSCWOA-ELM model, of which the root mean square error (RMSE) is 0.02341 and the mean absolute percentage error (MAPE) is 1.82175%, shows a better performance in predicting the ice coating damage to power line.