Application of artificial intelligence algorithms in predicting tunnel convergence to avoid TBM jamming phenomenon

One of the most important issues in TBM excavated tunnels is the exact estimation of the ground squeezing. Prediction of the ground behavior ahead of the tunnel face is essential to avoid project setbacks such as jamming phenomenon due to squeezing conditions. Artificial intelligence (AI) algorithms are proved to be suitable tools when relationship between dependent and independent variables cannot easily be understood. In this paper, well-known AI based methods, support vector machines (SVM) and artificial neural networks (ANN), were employed to predict ground condition of a tunneling project. The Ghomroud water conveyance tunnel excavated in rocks vulnerable to squeezing condition was selected as the case study. Training of the AI models was performed using previous practical experiences in the form of database. The tunnel convergence due to squeezing was considered as the models' outputs. According to the obtained results, it was observed that AI based methods can effectively be implemented for prediction of rock conditions in the tunneling projects. Moreover, it was concluded that performance of the SVM model is better than the ANN model. A high conformity was observed between predicted and measured convergence for the SVM model.

► Application of AI algorithms in predicting tunnel convergence is investigated.
► Two AI based models namely SVR and MLP are adopted to be implemented.
► The tunnel convergence due to squeezing was considered as the models' outputs.
► Ghomroud tunnel having convergence problem is selected for testing the AI models.