SSFLNet: A Novel Fault Diagnosis Method for Double Shield TBM Tool System

Area of Science:

• Engineering
• Mechanical Engineering
• Predictive Maintenance

Background:

• Tooling system wear in tunnel boring machines (TBMs) significantly impacts efficiency, costs, and safety.
• Detecting disk cutter wear during operation is challenging, leading to potential project delays and increased expenses.

Purpose of the Study:

• To propose a novel fault diagnosis method for TBM tooling systems, specifically addressing disk cutter wear.
• To enhance the detection of wear-related faults, thereby improving TBM operational efficiency and safety.

Main Methods:

• Developed a 3D model of the TBM hydraulic thrust and tool systems using SolidWorks.
• Conducted dynamic simulations with Adams to analyze hydraulic cylinder load variations due to cutter wear.
• Modeled and simulated the hydraulic propulsion system in AMESIM, acquiring pressure and flow signals.
• Established a SAV-SVDD failure location (SSFL) network model for fault diagnosis.

Main Results:

• The SSFL network model achieved 90% accuracy in identifying the failure area of the cutter head.
• Correlations were established between hydraulic cylinder signals (pressure, flow) and the extent of tooling system failure.
• The SSFL model demonstrated superior performance compared to SAE-SVM and SVDD models.

Conclusions:

• The proposed SSFL method effectively diagnoses disk cutter wear in TBMs.
• Timely identification of worn tooling enables proactive replacement, mitigating efficiency losses.
• The study validates the feasibility of the SSFL approach for real-world TBM applications.