Noise-robust self-supervised learning with frequency-bias decomposition for TBM muck particle size distribution prediction

Key Points

• Prediction accuracy of muck particle size distribution significantly improved with novel noise-robust methods.
• An experimental evaluation showed that self-supervised learning approaches effectively reduced noise influences.
• Utilizing frequency-bias decomposition enhanced predictive capabilities for various muck sizes and types.
• The findings indicate potential advancements in tunneling and excavation projects, highlighting the need for robust models.