Fig 1.
Model of double-circuit transmission line on the same tower.
Table 1.
9 single disturbance and some composite disturbances.
Fig 2.
One-dimensional global attention mechanism (1D-GAM) module.
Fig 3.
MGCNN structure.
Table 2.
MGCNN model specific parameters.
Fig 4.
SDTransformer base model.
Fig 5.
PQDs recognition framework based on MGCNN-SDTransformer.
Table 3.
Training environment parameters.
Fig 6.
Model training process.
Table 4.
29 Categories of recognition accuracy.
Table 5.
Comparison of model recognition accuracy.
Table 6.
Comparison of different levels of global convolution modules.
Table 7.
Comparison of different SDTransformer network depths.
Table 8.
Comparison of the number of MSA in SDTransformer network.
Table 9.
Ablation experiments between different modules.
Fig 7.
Recognition accuracy of 29 disturbance types under different networks.
Fig 8.
Disturbance integrated recognition rate under different networks.
Table 10.
Comparison of recognition rates of different models.
Fig 9.
Recognition results of real dataset.
Table 11.
Recognition results of real dataset.