Research on predicting the permeability of sintered material layer based on improved VMD-CNN-GRU model

To address the issues of insufficient accuracy and poor real-time performance in sinter bed permeability prediction, this paper proposes a prediction model based on an improved variational mode decomposition (VMD) and convolutional neural network-gated recurrent unit (CNN-GRU) hybrid approach. The model employs VMD combined with wavelet threshold denoising to preprocess the raw data, decomposing the complex sequence into multiple intrinsic mode functions (IMFs) with distinct frequency characteristics, followed by signal reconstruction using wavelet thresholding. Based on this, a CNN-GRU hybrid prediction model is constructed by leveraging the spatial feature extraction capability of CNN and the temporal modeling advantage of GRU. After model training and hyperparameter optimization, experimental results show that the VMD-CNN-GRU model outperforms the comparison models in MAPE, RMSE, and MAE metrics, achieving an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value of 96%, demonstrating superior prediction accuracy and stability. This provides an effective forecasting method for controlling and optimizing sinter bed permeability.