Reliable monitoring and prediction method for transmission lines based on FBG and LSTM

Transmission lines are susceptible to extreme weather conditions, and severe icing disasters can lead to incidents such as line breakage and collapse. Traditional monitoring and prediction methods for managing ice disasters suffer from poor reliability and short prediction lead times, hindering effective disaster prevention and mitigation efforts. This study introduces a prediction system enhancing icing forecast accuracy and timing. Initially, a dependable architecture was developed for gathering microclimate data on transmission lines using fiber Bragg grating technology. Subsequently, an optimized icing prediction process was established. The Bayesian optimization algorithm was utilized to optimize the entire predictive process, from input through the internal structure of the model to the final output, enhancing the accuracy and reliability. The prediction outcomes of various models, including recurrent neural networks, long short-term memory, gated recurrent units, and artificial neural networks, were then compared across different time series settings. The optimal prediction model was validated across three icing cycles collected in different provinces, achieving icing forecasts 6 hours in advance. With an R-squared value exceeding 0.97 and a mean absolute percentage error below 1.5%, the model demonstrated versatility under various conditions. This method, by outperforming current prediction techniques, significantly enhances forecasting precision and duration, effectively elevating the level of ice disaster prevention and control.