Transmission Line Meteorological Risk Classification Using Multi-Source Incomplete Data

In this paper, an innovative meteorological risk classification method is proposed to solve the problem of risk prediction of transmission lines under the complex and changeable meteorological data environment. This method is based on the incomplete meteorological conditions and risk label data sets of multiple regions, and uses the domain adaptive technology to process these incomplete and different domain data. By constructing a data completion module, a multi-source domain adaptive module based on an anchoring adapter and a classifier based on feature certainty, the meteorological risk prediction effect of the transmission dense line is improved, and especially the analysis capability of an unknown situation which does not occur in history is improved.