Research on Power OPGW Cable Operating State Early-Warning Method based on CNN-LSTM Prediction

Aiming at the problem of insufficient robustness of the current power OPGW cable early-warning mechanism, this paper proposed a power OPGW cable operating state early-warning method based on CNN-LSTM prediction. Firstly, the data prediction model of power OPGW cable is established based on CNN-LSTM training network to provide the data basis for further early-warning operation, in which the prediction value of OPGW fiber-core strain data-sequence is obtained by means of CNN feature extraction and LSTM data prediction. Then, MLP classification identification model is utilized to make the decision of operating state early-warning for power OPGW cable, in which the prediction data of OPGW fiber core strain is used as the data source. Finally, the experiment is carried out to verify the effectiveness of the proposed method. The experimental results illustrate that, compared with the LSTM-prediction method, the proposed method can improve the prediction MAE of OPGW fiber-core strain data-sequence from 0.046 to 0.038, and the average accuracy rate of early-warning classification reached 96.99%, achieving the effectiveness for the better operation and maintenance of power OPGW cable more efficiently.