Short-Term Prediction for Transmission Lines Icing Based on BP Neural Network

This paper proposes a short-term prediction model for transmission lines icing based on back- propagation (BP) neural network. Our work begins with a review of basic principles of this model and a brief research background. Then, a series of preprocessing on the data obtained from on-line icing monitoring system, such as the elimination of data with crassitude error, the correction of anomaly data, the supplementation of missing data and the normalization of data, are carried out. Finally, tests are conducted to evaluate the validity and the applicability of the proposed predicting model by using field data provided by Early-Warning System for Transmission Lines Disaster (Icing) of China Southern Power Grid. Results have shown that the proposed model based on three-layer BP neural network is fairly accurate for the short- term prediction of transmission lines icing in different areas.