Risk early warning method of severe convective disasters for transmission lines based on radar echo and LSTM

Abstract

In severe convective weather, transmission lines are prone to lightning strikes, wind swings, rain flashes and other faults that threaten the safe operation of the power grids. To overcome the problem that the existing nowcasting cannot fully meet the demand for refined weather forecasts for transmission line risk warning, a deep-learning-based nowcasting model of severe convective is constructed in this paper. The model is built by using meteorological radar echo image, assimilated data of wind speed and rainfall, lightning location data of the power grid and it is used to carry out early warning of transmission line risks. First, an long short-term memory (LSTM) network-based forecasting model of severe convective meteorological elements is constructed by taking the meteorological radar echo and its time-series extrapolation data as inputs, and assimilated data of wind speed and rainfall, lightning density, lightning current intensity as outputs. Then, combined with the wind, rain, and lightning forecast output from the model, the risk of wind swing, lightning strike, and tower collapse in the transmission corridor grid is evaluated. The fault probability of the transmission line is calculated in a comprehensive way for risk early warning. Finally, a case of successful early warning of severe convective weather processes in a province in September 2023 is demonstrated, which demonstrates the ability of the proposed method to improve the risk early warning capability of transmission lines under severe convective weather.

Keywords

• transmission lines
• severe convective disasters
• risk early warning
• radar echo
• long short-term memory (lstm) network
• wind swing discharge