Multi-sensor microseismic event classification with SNR-based sensor selection and MPA-SincNet model

The uploaded .zip file contains data and network of mircoseismic classification program. network Folder: get_dataset.py: Used to construct the dataset. ippa.py: Implements the IPPA attention mechanism, which consists of an Adaptive Average Pooling (AAP) block and an Adaptive Max Pooling (AMP) block. sincnet.py: Contains the first layer structure of the SincNet model. model.py: Implements the overall structure of the MPA-SincNet model, combining SincNet and the IPPA attention mechanism. resources Folder: checkpoint.pth: Saves model parameters during training for resuming training or inference. raw_train_data.pkl: Raw training data, consisting of unprocessed signal data. signal_selected_predict_data.pkl: Prediction data after signal selection and preprocessing. signal_selected_train_data.pkl: Training data after signal selection and preprocessing. main Folder: data_select_and_preprocess.py: Used for signal selection and data preprocessing. train.py: Contains training and validation code, responsible for the model training process, including loss calculation, optimizer setup, model saving, and performance evaluation. predict.py: Contains prediction code, used to make predictions on new data using the trained model. sensor_info Folder: Contains sensor coordinates and detailed information on the event snr selected for each of the four sensor selection methods.