Intelligent decision method for stability assessment of shield tunnel based on multi-objective data mining.

Area of Science:

• Geotechnical Engineering
• Tunneling Engineering
• Artificial Intelligence in Infrastructure

Background:

• Shield construction in complex strata poses significant risks, particularly excavation face instability.
• Accurate assessment of excavation face stability is critical for preventing catastrophic accidents in underground projects.

Purpose of the Study:

• To establish an analytical formula for limit support pressure in soil during shield tunneling.
• To quantitatively assess excavation face stability using a support safety coefficient.
• To develop a reliable evaluation model for underwater shield tunnel excavation face stability.

Main Methods:

• Developed an analytical formula for limit support pressure using a wedge model.
• Employed the rough set algorithm to reduce evaluation indices for excavation face stability.
• Utilized a back propagation (BP) neural network to create a predictive evaluation model.
• Validated the BP model's performance against TOPSIS and cloud models.

Main Results:

• Established a quantitative method for assessing excavation face stability.
• Developed a BP neural network model with a low prediction error (5.7675 × 10^-4).
• Demonstrated the BP model's superior prediction performance compared to other models.

Conclusions:

• The proposed evaluation method offers a vital reference for assessing underwater shield tunnel excavation face stability.
• The integration of analytical formulas and AI models enhances the safety and reliability of shield construction.
• This research contributes to the field of failure analysis in transportation infrastructure using artificial intelligence.