Wind-induced response prediction of coupled towers and lines of a transmission tower-line system based on an LSTM network

Abstract

Transmission tower-line systems are highly susceptible to wind loads. This study proposes a long short-term memory (LSTM)-based method to predict the wind-induced dynamic response of a transmission tower-line system. First, an LSTM model is established to predict the wind-induced nonlinear response of an individual transmission tower via wind tunnel testing data. The predicted results agree well with the measurements under various wind speeds. Despite the noise in the measurements, the LSTM model can capture the dynamic response characteristics. Second, the developed LSTM-based model is refined to predict the dynamic response of a transmission tower-line system under various wind speeds. The key novelty of the proposed method is that it can accurately predict the wind-induced nonlinear responses of transmission lines via the response of transmission towers. It is found that the proposed LSTM-based model can effectively capture the nonlinear relationship between the transmission tower and the transmission line. Quantitative analysis indicates that the overall prediction accuracy exceeds 90 %, which validates the method’s accuracy and generalization capability under different conditions. The present study offers an efficient and accurate LSTM-based method to predict the complete dynamic responses of transmission tower-line systems via limited measurements.