Distributed OPGW abnormal vibration monitoring and forewarning based on LSTM

Analyzing and predicting abnormal vibrations in optical fiber composite overhead ground wire (OPGW) transmission lines accurately is a challenging task. This paper proposes a distributed monitoring and forewarning method for OPGW abnormal vibrations using the long short-term memory (LSTM) algorithm, leveraging the regularity of abnormal vibrations related to climatic conditions. A distributed fiber Bragg grating array is employed to acquire monitoring signals, followed by the derivation of LSTM prediction steps. We effectively capture the long-term dependence of OPGW abnormal vibration signals by introducing cell state and gating mechanisms. In addition, the abnormal vibration forewarning method is analyzed by correlating predicted data with historical data. Experimental results in Hulunbuir demonstrate that the LSTM algorithm performs well in predictions over a 22-h period, evidenced by a root mean square error of 0.8729 and a determination coefficient (R2) of 0.9938 for the fitting curve with actual results. This performance surpasses that of the traditional GA-BP algorithm, facilitating effective abnormal vibration forewarning. This method holds significant potential for widespread application in the field of OPGW abnormal vibration engineering.