Utilizing meta-heuristic algorithms for load-bearing capacity prediction in piles with support vector regression

When designing foundations and using geotechnical engineering, pile bearing capacity (Pu) is essential, indicating the maximum load piles can sustain without failure. Accurate Pu calculation ensures structural stability and safety, considering soil conditions and structural loads. Methods range from...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Multiscale and Multidisciplinary Modeling, Experiments and Design Ročník 7; číslo 6; s. 5445 - 5459
Hlavní autoři: Xu, Mengran, Zhu, Zhihui
Médium: Journal Article
Jazyk:angličtina
Vydáno: Cham Springer International Publishing 01.11.2024
Témata:
Characterization and Evaluation of Materials
Engineering
Mathematical Applications in the Physical Sciences
Mechanical Engineering
Numerical and Computational Physics
Original Paper
Simulation
Solid Mechanics
Support vector regression
Machine-learning
Population-based Vortex Search Algorithm
Electric Charged Particles Optimization
Pile bearing capacity
ISSN:2520-8160, 2520-8179
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:When designing foundations and using geotechnical engineering, pile bearing capacity (Pu) is essential, indicating the maximum load piles can sustain without failure. Accurate Pu calculation ensures structural stability and safety, considering soil conditions and structural loads. Methods range from empirical equations to advanced numerical analyses, factoring in soil characteristics and pile dimensions. Recent innovations like machine learning enhance prediction accuracy. Understanding P u is vital for optimizing pile designs, reducing risks, and promoting resilient, sustainable construction. Advances in P u prediction promise continued improvements in construction practices. This research employs the support vector regression (SVR) model as a key problem-solving approach in developing a robust machine-learning framework. To improve the accuracy and performance of the model, it integrates two distinct meta-heuristic optimization techniques: the Population-based Vortex Search Algorithm (PVSA) and the Electric Charged Particles Optimization (ECPO). These optimization methods are strategically harnessed to fine-tune the SVR model parameters, ensuring the attainment of optimal outcomes. By leveraging ECPO and PVSA, the research aims to push the boundaries of predictive accuracy and computational efficiency in machine-learning applications. Based on the results obtained, it became apparent that the SVPV (SVR + PVSA) model, which amalgamates the SVR model with the PVSA optimization technique, yielded the most precise estimations for P u values. This conclusion was supported by an 0.995 R 2 and an 31.305 RMSE value, both of which underscored the model's exceptional predictive capabilities.
ISSN:2520-8160
2520-8179
DOI:10.1007/s41939-024-00527-y