Fig 1.
LSTM-CNN Feature extractor.
Fig 2.
Flowchart of the GSSSA optimization framework.
Fig 3.
Block diagram for identifying external force damage sources vibration signals.
Table 1.
Experimental software and hardware configuration.
Fig 4.
Distribution map of t-SNE characteristics.
(a) LSTM features (b) CNN features (c) LSTM-CNN fusion features.
Table 2.
Comparison of classification performance indicators of recognition models.
Fig 5.
(a) LSTM-CatBoost (b) CNN-CatBoost (c) LSTM-CNN-CatBoost (d) LSTM-CNN-CatBoost-GSSSA.
Table 3.
Evaluation results of LSTM-CatBoost.
Table 4.
Evaluation results of CNN-CatBoost.
Table 5.
Evaluation results of LSTM-CNN-CatBoost.
Table 6.
Evaluation results of LSTM-CNN-CatBoost-GSSSA.
Fig 6.
Bar line chart of accuracy-training time.
Fig 7.
Comparison chart of the recognition accuracy of seven types of algorithms.