Figure 5
The figure shows a single line chart titled “Training vs Validation Loss”. The horizontal axis is labeled “Epoch” and displays values from 0 to 30, increasing from left to right with an interval of 5. The vertical axis is labeled “Loss” and ranges from 0 at the bottom to 500 at the top, with evenly spaced tick marks at an interval of 100. The legend is positioned in the upper right corner. Two solid lines represent “Train Class Loss” and “Val Class Loss”. Two dashed lines represent “Train R U L Loss” and “Val R U L Loss”. The dashed Train R U L Loss line starts near 500 at epoch 0 and decreases sharply over the first few epochs, dropping below 200 by around epoch 2 and near 100 by about epoch 5. It continues to decline more gradually, fluctuating between approximately 20 and 50 from around epoch 10 through epoch 30. The dashed Val R U L Loss line starts near 350 at epoch 0, decreases rapidly during the early epochs, and falls below 150 by about epoch 4. Afterward, it continues to decrease with noticeable fluctuations, remaining mostly between approximately 20 and 60 from around epoch 10 to epoch 30. The solid Val Class Loss line appears very close to zero across all epochs, forming a nearly flat line just above the horizontal axis. Light grid lines are visible across the plot, aligned with both axes. Note: All numerical data values are approximated.Training and validation loss profiles of the CNN–LSTM digital twin during optimization. The gradual and monotonic decrease in both loss curves demonstrates effective gradient propagation and model stability. The close alignment between training and validation losses confirms that the model generalizes well without evidence of underfitting or overfitting