TBM disc cutter wear prediction using stratal slicing and IPSO-LSTM in mixed weathered granite stratum
•Propose a stratal slicing method to reflect the actual cutter working conditions.•The stratal slicing method has stronger geological information processing ability.•It can improve the prediction accuracy and has real-time prediction capability.•The error of combined method in real-time predictions...
Gespeichert in:
Veröffentlicht in: | Tunnelling and underground space technology 2024-06, Vol.148, p.105745, Article 105745 |
---|---|
Hauptverfasser: | , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | •Propose a stratal slicing method to reflect the actual cutter working conditions.•The stratal slicing method has stronger geological information processing ability.•It can improve the prediction accuracy and has real-time prediction capability.•The error of combined method in real-time predictions is close to that of the manual measure.•It is expected to demonstrate excellent performance in more complex geological conditions and larger shield diameters.
Monitoring the wear status of cutters is important for safe and sustainable shield construction and cost management. In this paper, an innovative stratal slicing method is proposed to convert segmented and discrete uniaxial compressive strength (UCS) test data into a sequential dataset by combining it with a geological profile. The sequential dataset not only accurately represents the changing strata conditions but also differentiates the working conditions of the disc cutters in various cutterhead areas on the excavation face. Its sequence characteristics can be better combined with shield operational parameters and time-series models for real-time prediction. Furthermore, the particle swarm optimization (PSO) algorithm was improved by adding variable inertia weights and elimination mechanisms, which effectively optimised the hyperparameters of the long short-term memory (LSTM) model. The proposed method was applied to a field tunnelling case with data collected from the Guangzhou Metro Line 18 railway. The results show that the UCS data obtained using the proposed stratal slicing method can improve the prediction accuracy compared to traditional methods and models. In particular, the combined IPSO + LSTM + horizontal summation method can obtain the most accurate prediction results and has real-time prediction capability. With the proposed stratal slicing method, the IPSO + LSTM modelling approach is generally applicable to more complex ground conditions and larger shield diameters. |
---|---|
ISSN: | 0886-7798 1878-4364 |
DOI: | 10.1016/j.tust.2024.105745 |