Ground Settlement Prediction in Urban Tunnelling: Leveraging Metaheuristic-Optimized Random Forest Models DOI Creative Commons
Peixi Yang, Jian Zhou, Yulin Zhang

et al.

Arabian Journal for Science and Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: March 5, 2025

Abstract With the continuous acceleration of urbanization, problem ground settlement induced by underground tunnel construction has received more and widespread attention. This study addresses challenge predicting surface subsidence in urban construction, a critical concern geotechnical engineering. Random forest (RF) models were optimized using three distinct metaheuristic algorithms: ant lion optimizer (ALO), multiverse (MVO), grasshopper optimization algorithm (GOA). The enhancements significantly improved model accuracy, as demonstrated detailed performance metrics GOA-optimized RF (GOA-RF Pop = 20) on Changsha Metro Line 3 dataset, which included 294 instances 12 feature parameters. achieved an MAE 1.3820, MAPE 181.2249, correlation coefficient 0.9273, RMSE 2.5209 training set; 2.4695, 275.2054, R value 0.8877, 4.2540 testing set. A sensitivity analysis within random framework revealed that torque (To) condition (Gc) had most significant impact subsidence, whereas influence modified dynamic penetration test (MDPT) was least pronounced. Additionally, MATLAB-based application developed App Designer module, integrating these into user-friendly GUI facilitates prediction management risks, thereby enhancing practical effectiveness engineering risk mitigation strategies.

Language: Английский

Ground Settlement Prediction in Urban Tunnelling: Leveraging Metaheuristic-Optimized Random Forest Models DOI Creative Commons
Peixi Yang, Jian Zhou, Yulin Zhang

et al.

Arabian Journal for Science and Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: March 5, 2025

Abstract With the continuous acceleration of urbanization, problem ground settlement induced by underground tunnel construction has received more and widespread attention. This study addresses challenge predicting surface subsidence in urban construction, a critical concern geotechnical engineering. Random forest (RF) models were optimized using three distinct metaheuristic algorithms: ant lion optimizer (ALO), multiverse (MVO), grasshopper optimization algorithm (GOA). The enhancements significantly improved model accuracy, as demonstrated detailed performance metrics GOA-optimized RF (GOA-RF Pop = 20) on Changsha Metro Line 3 dataset, which included 294 instances 12 feature parameters. achieved an MAE 1.3820, MAPE 181.2249, correlation coefficient 0.9273, RMSE 2.5209 training set; 2.4695, 275.2054, R value 0.8877, 4.2540 testing set. A sensitivity analysis within random framework revealed that torque (To) condition (Gc) had most significant impact subsidence, whereas influence modified dynamic penetration test (MDPT) was least pronounced. Additionally, MATLAB-based application developed App Designer module, integrating these into user-friendly GUI facilitates prediction management risks, thereby enhancing practical effectiveness engineering risk mitigation strategies.

Language: Английский

Citations

0