Multi-Source DC Transmission Lines Icing Prediction Model Based on Adaptive Weight Feature

This paper analyzes and studies the relationship between line ice cover and its influencing factors by collecting and organizing transmission line ice cover-micro-meteorological monitoring data, meteorological automatic station monitoring data and ECMWF numerical prediction data of Xinjiang Power Grid, establishes an ice cover prediction model by BP neural network algorithm for the three types of data respectively, and finally performs dynamic fusion prediction on the results of the multi-source ice cover prediction. The main innovation of this paper is that a dynamic fusion method of multi-source ice cover prediction with adaptive weight features is adopted to fuse the multi-source ice cover prediction results, which improves the accuracy and reliability of the prediction by combining the measured data and numerical prediction data.