fig4

Figure 4. Transformer-CNN-BiGRU Model structure. (A) Model framework. Schematic diagram of (B) CNN structure, (C) Transformer structure, (D) BiGRU structure. CNN: Convolutional neural network; BiGRU: bidirectional gated recurrent unit; CEEMDAN: complementary ensemble empirical mode decomposition with adaptive noise; 1D: one-dimensional; IMF: intrinsic mode function; GRU: gated recurrent unit; CNN: convolutional neural network; HI: health indicator.