Figures (10)  Tables (1)

    • Figure 1. 

      Experimental procedure.

    • Figure 2. 

      Geolocation of the observation sites.

    • Figure 3. 

      The NH3000 automatic wire icing monitoring system and its installation setup.

    • Figure 4. 

      (a)−(e) Original and (f)−(j) cropped images of wire icing for different risk levels (Levels 0−4).

    • Figure 5. 

      (a) Cross-Entropy Loss curves and (b) accuracy curves for the training and validation sets of VGG16 and ResNet34 models.

    • Figure 6. 

      (a) Cross-Entropy Loss curves and (b) accuracy curves for the training and validation sets of ResNet34 and ResNet34+ models.

    • Figure 7. 

      Confusion matrix of the recognition accuracy for different wire icing risk levels using the (a) ResNet34 and (b) ResNet34+ model.

    • Figure 8. 

      Confusion matrices of the recognition accuracy for different wire icing risk levels using the ResNet34+ model in various regions and different directions of wires: (a) Enshi East-West direction, (b) Enshi North-South direction, (c) Shennongjia East-West direction, and (d) Shennongjia North-South direction.

    • Figure 9. 

      The accuracy curves for different time intervals in wire icing risk level identification.

    • Figure 10. 

      Confusion matrices of the recognition accuracy for different wire icing risk levels using the ResNet34+ model in different directions of wires (a), (b) and the accuracy curves for different time intervals in ice accumulation risk level identification (c), (d) in Mount. Lu.

    • Model hyperparameters Specific values/algorithms
      Batch size 128
      Max epochs 100
      Patience 15
      Learning rate 0.001
      Random horizontal flip probability 0.3
      Random rotation probability 0.3
      Random rotation Angle range 30°
      Random image filtering probability 0.2
      Optimizer Adam
      Dropout probability 0.2

      Table 1. 

      Specific values or algorithms for adjusted model hyperparameters.