Abstract: Safe and efficient control of hard rock tunneling by Tunnel Boring Machines is a core issue in the intelligent construction of coal mine rock roadways. Accurate prediction of TBM excavation parameters serves as a critical prerequisite for achieving safe, efficient, and intelligent construction, holding significant importance for reducing construction risks and optimizing tunneling performance. In response to the limitations of existing prediction methods in areas such as mining temporal correlation features, hyperparameter optimization, and model interpretability, this paper proposes an OBE−Net deep learning prediction model that integrates automatic hyperparameter optimization and an attention mechanism. The model employs a Bidirectional Long Short−Term Memory network to capture the temporal dependencies of the tunneling parameters, introduces an Efficient Channel Attention mechanism to achieve adaptive optimization of feature weights, and utilizes the Optuna framework for automatic global hyperparameter tuning, with the aim of achieving high−precision prediction of TBM tunneling parameters in coal mines. The research is based on engineering measurement data from a TBM−excavated rock roadway in a Huainan coal mine, with the total thrust force as the prediction target for model validation. The results indicate that the OBE−Net model achieved excellent predictive performance on the test set, with an R² of 0.9756, an RMSE of 0.0437, and an MAE of 0.0336. Its performance is significantly superior to baseline models such as standalone BiLSTM and GRU. Furthermore, the model demonstrated good generalization capability across data from different engineering sections. Interpretability analysis based on the SHAP framework reveals the contribution and degree of influence of each input parameter on the prediction results. The trained OBE−Net model is encapsulated into a standalone executable program, laying a foundation for the practical application of intelligent TBM tunneling parameter optimization technology in coal mines. It also provides methodological reference and a theoretical basis for tunneling parameter prediction in similar projects.