Semantic fusion and structured attention for intelligent simulation and optimization in underground excavation
Abstract
Intelligent control of complex cyber–physical systems requires integrating heterogeneous data such as sensor streams, control signals, and domain knowledge. In tunnel excavation, existing models often lack accuracy, adaptability, and transparency. This paper presents a transparent and structure-aware simulation–optimization framework for TBM operations by integrating heterogeneous operational data and semantic features. To enable explicit and domain-consistent integration of geological and contextual information, by incorporating concept-level embeddings, the framework integrates geological and contextual information while maintaining transparency and engineering consistency. Validated on the Yinsong Water Diversion Project in Jilin Province, China, the simulator achieved high predictive accuracy. Importantly, it advances automation in construction by enabling transparent data-driven optimization of excavation operations.
Details
- Organisation(s)
-
Institute of Photonics
- External Organisation(s)
-
Tongji University
- Type
- Article
- Journal
- Automation in construction
- Volume
- 189
- ISSN
- 0926-5805
- Publication date
- 26.05.2026
- Publication status
- E-pub ahead of print
- Peer reviewed
- Yes
- ASJC Scopus subject areas
- Control and Systems Engineering, Civil and Structural Engineering, Building and Construction
- Electronic version(s)
-
https://doi.org/10.1016/j.autcon.2026.107050 (Access:
Closed
)
-
Details in the research portal "Research@Leibniz University"
