ANN-based swarm intelligence for predicting expansive soil swell pressure and compression strength

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: June 25, 2024

Abstract This research suggests a robust integration of artificial neural networks (ANN) for predicting swell pressure and the unconfined compression strength expansive soils ( P s UCS -ES). Four novel ANN-based models, namely ANN-PSO (i.e., particle swarm optimization), ANN-GWO grey wolf ANN-SMA slime mould algorithm) alongside ANN-MPA marine predators’ were deployed to assess -ES. The models trained using nine most influential parameters affecting -ES, collected from broader range 145 published papers. observed results compared with predictions made by metaheuristics models. efficacy all these formulated was evaluated utilizing mean absolute error (MAE), Nash–Sutcliffe (NS) efficiency, performance index ρ , regression coefficient R 2 ), root square (RMSE), ratio RMSE standard deviation actual observations (RSR), variance account (VAF), Willmott’s agreement (WI), weighted percentage (WMAPE). All developed -ES had an significantly > 0.8 overall dataset. However, excelled in yielding high values training dataset TrD testing TsD validation VdD ). model also exhibited lowest MAE 5.63%, 5.68%, 5.48% respectively. model’s revealed that exceeded 0.9 . decreased Also, yielded higher (0.89, 0.93, 0.94) comparatively low (5.11%, 5.67, 3.61%) case PSO, GWO, SMA, witnessed overfitting problem because aforementioned 0.62, 0.56, 0.58 On contrary, no significant observation recorded ANN-base tested a-20 index. For maximum points lie within ± 20% error. sensitivity as well monotonicity analyses depicted trending corroborate existing literature. Therefore, it can be inferred recently built swarm-based ANN particularly ANN-MPA, solve complexities tuning hyperparameters ANN-predicted replicated practical scenarios geoenvironmental engineering.

Language: Английский

---

Synergistic evolution of hydrological and movement characteristics of Majiagou landslide and identification of key triggering factors through interpretable machine learning

Bulletin of Engineering Geology and the Environment, Journal Year: 2025, Volume and Issue: 84(2)

Published: Jan. 28, 2025

Language: Английский

---

Geotechnical Characterization and Stability Prediction of Nano-Silica-Stabilized Slopes: A Machine Learning Approach to Mitigating Geological Hazards

Transportation Infrastructure Geotechnology, Journal Year: 2025, Volume and Issue: 12(2)

Published: Feb. 1, 2025

Language: Английский

---

PSInSAR-Based Time-Series Coastal Deformation Estimation Using Sentinel-1 Data

Land, Journal Year: 2025, Volume and Issue: 14(3), P. 536 - 536

Published: March 4, 2025

Coastal areas are highly dynamic regions where surface deformation due to natural and anthropogenic activities poses significant challenges. Synthetic Aperture Radar (SAR) interferometry techniques, such as Persistent Scatterer Interferometry (PSInSAR), provide advanced capabilities monitor with high precision. This study applies PSInSAR techniques estimate along coastal zones from 2017 2020 using Sentinel-1 data. In the densely populated of Pasni, an annual subsidence rate 130 mm is observed, while northern, less region experiences uplift 70 per year. Seawater intrusion emerging issue causing in Pasni’s areas. It infiltrates freshwater aquifers, primarily excessive groundwater extraction rising sea levels. Over time, seawater destabilizes underlying soil rock structures, leading or gradual sinking ground surface. form risks infrastructure, agriculture, local ecosystem. Land varies area’s coastline. The eastern region, which reclaimed, particularly affected by erosion. results derived SAR data indicate major urban districts. information crucial for management, hazard assessment, planning sustainable development region.

Language: Английский

---

Application of Soft Computing Techniques for Slope Stability Analysis

Transportation Infrastructure Geotechnology, Journal Year: 2024, Volume and Issue: 11(6), P. 3903 - 3940

Published: Aug. 5, 2024

Language: Английский

---

A physics-informed machine learning solution for landslide susceptibility mapping based on three-dimensional slope stability evaluation

Journal of Central South University, Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 5, 2024

Language: Английский

---

A hybrid data-driven approach for rainfall-induced landslide susceptibility mapping: Physically-based probabilistic model with convolutional neural network

Journal of Rock Mechanics and Geotechnical Engineering, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 1, 2024

Language: Английский

---

Bayesian ensemble learning and Shapley additive explanations for fast estimation of slope stability with a physics-informed database

Natural Hazards, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 19, 2024

Language: Английский

---

Impact of sampling for landslide susceptibility assessment using interpretable machine learning models

Bulletin of Engineering Geology and the Environment, Journal Year: 2024, Volume and Issue: 83(11)

Published: Oct. 25, 2024

Language: Английский

---

Indirect Evaluation of the Influence of Rock Boulders in Blasting to the Geohazard: Unearthing Geologic Insights Fused with Tree Seed based LSTM Algorithm

Geohazard Mechanics, Journal Year: 2024, Volume and Issue: unknown

Published: June 1, 2024

Effective control of blasting outcomes depends on a thorough understanding rock geology and the integration geological characteristics with blast design parameters. This study underscores importance adapting parameters to conditions optimize utilization explosive energy for fragmentation. To achieve this, data fifty geo-blast were collected used train machine learning algorithms. The objective was develop predictive models estimating oversize percentage, incorporating seven controlled components one uncontrollable index. employed combination hybrid long-short-term memory (LSTM), support vector regression, random forest Among these, LSTM model enhanced tree seed algorithm (LSTM-TSA) demonstrated highest prediction accuracy when handling large datasets. LSTM-TSA soft computing specifically leveraged various such as burden, spacing, stemming length, drill hole charge powder factor, joint set number. estimated percentage values these determined 0.7 m, 0.9 0.65 1.4 1.03 kg/m3, 35%, 2, respectively. Application resulted in significant 28.1% increase crusher's production rate, showcasing its effectiveness improving operations.

Language: Английский

---