Probabilistic machine learning for predicting desiccation cracks in clayey soils

Bulletin of Engineering Geology and the Environment, Год журнала: 2023, Номер 82(9)

Опубликована: Авг. 17, 2023

Язык: Английский

---

An in-depth comparative analysis of data-driven and classic regression models for scour depth prediction around cylindrical bridge piers

Applied Water Science, Год журнала: 2023, Номер 13(12)

Опубликована: Ноя. 7, 2023

Abstract The study focuses on the critical concern of designing secure and resilient bridge piers, especially regarding scour phenomena. Traditional equations for estimating depth are limited, often leading to inaccuracies. To address these shortcomings, modern data-driven models (DDMs) have emerged. This research conducts a comprehensive comparison involving DDMs, including support vector machine (SVM), gene expression programming (GEP), multilayer perceptron (MLP), gradient boosting trees (GBT) multivariate adaptive regression spline (MARS) models, against two predicting around cylindrical piers. Evaluation employs statistical indices, such as root-mean-square error (RMSE), coefficient determination ( R 2 ), mean average (MAE) normalized discrepancy ratio S (DDRmax) assess their predictive performance. A total 455 datasets from previous papers employed assessment. Dimensionless parameters Froude number $$\left( {Fr = \frac{U}{{\sqrt {gy} }}} \right)$$ Fr=Ugy , Pier $$Fr_{P} {g^{\prime } D} }}$$ xmlns:mml="http://www.w3.org/1998/Math/MathML">FrP=Ug′D pier diameter $$(\frac{\text{y}}{{\text{D}}})$$ xmlns:mml="http://www.w3.org/1998/Math/MathML">(yD) carefully selected influential model inputs through dimensional analysis gamma test. results highlight superior performance SVM model. In training phase, it exhibits an RMSE 0.1009, MAE 0.0726, 0.9401, DDR 2.9237. During testing, shows 0.023, 0.017, 0.984, 5.301. Additionally, has − 0.065 20.642 in set 0.005 0.707 testing set. Conversely, M5 lowest accuracy. metrics unequivocally establish significantly outperforming experimental placing higher echelon

Язык: Английский

---

A novel deep learning model to predict the soil nutrient levels (N, P, and K) in cabbage cultivation

Smart Agricultural Technology, Год журнала: 2023, Номер 7, С. 100395 - 100395

Опубликована: Дек. 31, 2023

Cabbage (Brassica oleracea) is a green cruciferous vegetable. Major nutrients (nitrogen, phosphorus, and potassium) are frequently applied to the soil due low fertility levels. However, optimizing required fertilizer levels extremely important avoid any overuse underuse. Therefore, it develop comprehensive methodology for evaluating major in soil. In this research, deep learning model was introduced predict nitrogen, potassium content of by analyzing growing characteristics plants, such as plant height, number leaves, average leaf area plant. To achieve this, cabbage plants were recorded weekly along with respective nearby After data trained using Levenberg–Marquardt algorithm tested different transfer functions logarithmic sigmoid, pure linear, tangent better predictions obtained through model. According Pearson correlation values, linear sigmoid showed higher ranging from 0.99 training, testing, validation, all points model, indicating strong relationship between actual predicted values. Mean Square Error function outperformed others, giving value 1.0813,

Язык: Английский

---

Machine learning models coupled with empirical mode decomposition for simulating monthly and yearly streamflows: a case study of three watersheds in Ontario, Canada

Engineering Applications of Computational Fluid Mechanics, Год журнала: 2023, Номер 17(1)

Опубликована: Авг. 21, 2023

This paper presents a novel approach for enhancing long-term runoff simulations through the integration of empirical mode decomposition (EMD) with four machine learning (ML) models: ensemble, support vector (SVM), convolutional neural networks (CNN), and artificial backpropagation (ANN-BP). The proposed methodology uses EMD to decompose precipitation temperature time-series into intrinsic functions, thereby revealing underlying data patterns. Subsequently, these components are incorporated ML models simulate time-series. effectiveness hybrid is evaluated using streamflow obtained from Grand, Winnipeg, Moosonee Rivers in Ontario, Canada. Four widely used performance indices, namely, correlation coefficient, root mean square error (RMSE), absolute relative error, Nash–Sutcliffe efficiency, employed assess models’ performance. results demonstrate that EMD-ML exhibit significantly superior compared standalone methods. During validation phase, EMD-Ensemble, EMD-SVM, EMD-CNN, EMD-ANN-BP notable reductions RMSEs monthly estimates Grand River, amounting 11%, 22%, 8%, 33%, respectively, their non-EMD counterparts. Additionally, improved yearly Winnipeg 54%, 0.08%, 6%, 4.5% respectively. To further enhance accuracy estimates, an SVM-recursive feature elimination technique select more appropriate dataset all study cases. research underscores potential integrating simulations. outcomes highlight models, demonstrating ability generating lower biases than These findings hold significant implications field computational fluid mechanics can contribute understanding hydrological processes.

Язык: Английский

---

Solitary wave-induced 2D seepage effects on sediment incipient motion

Applied Ocean Research, Год журнала: 2025, Номер 155, С. 104456 - 104456

Опубликована: Фев. 1, 2025

Язык: Английский

---