Evaluation of Landslide Susceptibility of Mangshan Mountain in Zhengzhou Based on GWO-1D CNN Model

Sustainability, Год журнала: 2024, Номер 16(12), С. 5086 - 5086

Опубликована: Июнь 14, 2024

The Mangshan Mountain is located in the south bank of Yellow River, which belongs to typical loess plateau. Landslide disasters occur frequently this region, so it urgent carry out evaluation landslide susceptibility. Therefore, study takes as research object, selects 13 factors through multicollinearity diagnostic, Pearson correlation coefficient, and random forest importance analysis, uses grey wolf optimizer (GWO) algorithm optimize initial weights one-dimensional convolutional neural network model (1D CNN), build a GWO-1D CNN results show that GWO can significantly improve accuracy 1D model. final reaches 0.903, accuracy, area under ROC curve, kappa coefficients increase by 0.091, 0.098, 0.187, respectively; percentage very low, medium, high, high susceptibility areas 40.2%, 23.6%, 14.1%, 12.9%, 9.2%. findings provide scientific basis for prevention control disaster expand application

Язык: Английский

---

Multi-objective optimization of tribological properties of camshaft bearing pairs using DNN coupled with NSGA-II algorithm and TOPSIS

Industrial Lubrication and Tribology, Год журнала: 2024, Номер 76(5), С. 703 - 715

Опубликована: Июнь 14, 2024

Purpose This study aims to propose an optimization framework using deep neural networks (DNN) coupled with nondominated sorting genetic algorithm II and technique for order preference by similarity ideal solution method improve the tribological properties of camshaft bearing pairs internal combustion engine. Design/methodology/approach A lubrication model based on theory elastohydrodynamic flexible multibody dynamics was developed a V6 diesel Setting DNN as fitness function, multi-objective decision-making were used optimize pair structure goal minimizing total friction loss difference average values minimum oil film thickness. Findings The results show that state corresponding optimized is lubrication. Compared original structure, significantly reduces loss. Originality/value performance structural parameters are obtained, verified through simulation. Peer review peer history this article available at: https://publons.com/publon/10.1108/ILT-12-2023-0417/

Язык: Английский

---

A body welded joint quality prediction method incorporating reconstruction error

Proceedings of the Institution of Mechanical Engineers Part C Journal of Mechanical Engineering Science, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 21, 2024

With the continuous development of artificial intelligence technology, application machine learning in prediction quality status body welded joints is becoming more and widespread. However, due to strict setting production parameters, proportion abnormal weld relatively small, resulting unbalanced data distribution. This poses a challenge accurately assess joint effectively identify joints. To address this problem, method incorporating reconstruction errors proposed, aiming predict In paper, noise-reducing autoencoder (DAE) used reconstruct feature data. It not only highlights key features, but also extracts error as new help model Because bigger. Then, features are inputs, score output. The support vector regression (SVR) particle swarm algorithm (PSO) construct mapping relationship between input output achieve status. experimental results show that highly reliable status, accuracy significantly better than other models. Compared with methods, identification 82%, 92%. study provides technical for intelligent online inspection automobile bodies.

Язык: Английский

---

A method based on improved ant colony algorithm feature selection combined with GWO-SVR model for predicting chlorophyll-a concentration in Wuliangsu Lake

Water Science & Technology, Год журнала: 2023, Номер 89(1), С. 20 - 37

Опубликована: Дек. 19, 2023

Abstract Chlorophyll-a (Chl-a) is an important parameter in water bodies. Due to the complexity of optics bodies, it difficult accurately predict Chl-a concentrations bodies by current traditional methods. In this paper, using Sentinel-2 remote sensing images as data source combined with measured data, taking Wuliangsu Lake study area, a new intelligent algorithm proposed for prediction concentration, which uses adaptive ant colony exhaustive optimization (A-ACEO) feature selection and gray wolf (GWO) optimize support vector regression (SVR) achieve concentration prediction. The improved select bands introducing relevant strategies. GWO-SVR model built optimizing SVR GWO selected input comparing model. results show that usage presented A-ACEO inputs can effectively reduce improve performance model, under condition same provide valuable references monitoring Lake.

Язык: Английский

---

A Predictive Model for the Freeze-Thaw Concrete Durability Index Utilizing the Deeplabv3+ Model with Machine Learning

Опубликована: Янв. 1, 2024

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Язык: Английский

---