A quantitative division method of deep roadway surrounding rock loose circle
By combining external anchor-type UWFBG strain optical cable and PSO-XGBoost regression model with numerical simulation dataset, quantitative division of loosened rock zones in deep roadways was achieved, solving the problems of insufficient monitoring accuracy and inconsistent standards in existing technologies, and providing a high-precision monitoring method for loosened rock zones.
Patent Information
- Application Number
- CN202611231064.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-14
- Publication Date
- 2026-09-25
AI Technical Summary
Existing technologies for delineating loosened zones in deep tunnels suffer from problems such as strong subjectivity, lack of unified standards, insufficient adaptability of sensing equipment, and inconsistent labeling standards for machine learning training data, resulting in monitoring accuracy that is difficult to meet engineering requirements.
The strain values of the surrounding rock were obtained by using an external anchor point UWFBG strain optical cable. Combined with the PSO-XGBoost regression model, a diverse training dataset was constructed through numerical simulation. The relative strain change rate was used as the input feature to achieve quantitative division of the loosened zone of the surrounding rock.
It enables effective monitoring of large deformations in the surrounding rock of deep roadways, solves the problems of insufficient range and poor long-term stability of sensing equipment, provides a unified judgment standard, and improves monitoring accuracy and cross-condition generalization ability.