Machine learning prediction and parameter sensitivity analysis of top earth pressure of high-filled cut-and-cover tunnels DOI
Sheng Li, Zi–Qiang Lang, Ke Zhang

et al.

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Abstract To quickly and effectively predict the unknown top earth pressure of high-filled cut-and-cover tunnels (HFCCTs), a prediction model HFCCTs based on machine learning was proposed. The data set established by taking Poisson 's ratio, friction angle, cohesion, ratio groove width to HFCCT width, slope angle filling height as input parameters ML models, maximum vertical output parameter, correlation between analyzed. Newton-Raphson-based optimization (NRBO) used optimize hyper-parameters XGBoost model, compared with XGBoost, SVM, RF, BP models under grid search. SHAP method analyze sensitivity NRBO-XGBoost model. Finally, field measured an airport high-speed railway tunnel, engineering applicability proposed verified. results revealed that among parameters, most influential factor. performance better than other traditional it has good applicability, which can provide effective basis for judging stability state in practical geotechnical engineering.

Language: Английский

Consistent African vulture optimization algorithm for electrical energy exchange in commercial buildings DOI
Linfei Yin, Jing Tian,

Xiaofang Chen

et al.

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 134741 - 134741

Published: Jan. 1, 2025

Language: Английский

Citations

1

Dynamic Deformation Analysis of Super High-Rise Buildings Based on GNSS and Accelerometer Fusion DOI Creative Commons
Xingxing Xiao, Houzeng Han, Jian Wang

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(9), P. 2659 - 2659

Published: April 23, 2025

To accurately capture the dynamic displacement of super-tall buildings under complex conditions, this study proposes a data fusion algorithm that integrates NRBO-FMD optimization with Adaptive Robust Kalman Filtering (ARKF). The method preprocesses GNSS and accelerometer to mitigate multipath effects, unmodeled errors, high-frequency noise in signals. Subsequently, ARKF fuses preprocessed achieve high-precision reconstruction. Numerical simulations varying conditions validated algorithm’s accuracy. Field experiments conducted on Hairong Square Building Changchun further demonstrated its effectiveness estimating three-dimensional displacement. Key findings are as follows: (1) significantly reduced while preserving essential signal characteristics. For data, root mean square error (RMSE) was 0.7 mm for 100 s dataset 1.0 200 dataset, corresponding signal-to-noise ratio (SNR) improvements 3.0 dB 6.0 dB. RMSE (100 s) 6.2 (200 s), 4.1 SNR gain. (2) NRBO-FMD–ARKF achieved high accuracy, values 1.9 s). Consistent PESD POSD long-term stability effective suppression irregular errors. (3) successfully fused 1 Hz overcoming limitations single-sensor approaches. yielded an 3.6 mm, 2.6 4.8 demonstrating both precision robustness. Spectral analysis revealed key response frequencies ranging from 0.003 0.314 Hz, facilitating natural frequency identification, structural stiffness tracking, early-stage performance assessment. This shows potential improving integration health monitoring. Future work will focus real-time predictive estimation enhance monitoring responsiveness early-warning capabilities.

Language: Английский

Citations

0

Machine learning prediction and parameter sensitivity analysis of top earth pressure of high-filled cut-and-cover tunnels DOI
Sheng Li, Zi–Qiang Lang, Ke Zhang

et al.

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Abstract To quickly and effectively predict the unknown top earth pressure of high-filled cut-and-cover tunnels (HFCCTs), a prediction model HFCCTs based on machine learning was proposed. The data set established by taking Poisson 's ratio, friction angle, cohesion, ratio groove width to HFCCT width, slope angle filling height as input parameters ML models, maximum vertical output parameter, correlation between analyzed. Newton-Raphson-based optimization (NRBO) used optimize hyper-parameters XGBoost model, compared with XGBoost, SVM, RF, BP models under grid search. SHAP method analyze sensitivity NRBO-XGBoost model. Finally, field measured an airport high-speed railway tunnel, engineering applicability proposed verified. results revealed that among parameters, most influential factor. performance better than other traditional it has good applicability, which can provide effective basis for judging stability state in practical geotechnical engineering.

Language: Английский

Citations

0