Determining International Irregularity Index (IRI) Values Through Artificial Neural Network (ANN) Modelling

Academic Platform Journal of Engineering and Smart Systems, Journal Year: 2025, Volume and Issue: 13(1), P. 7 - 16

Published: Jan. 31, 2025

The quality of a pavement's level service is generally determined by measuring the combinations some important factors which affect speed, travel time, freedom to maneuver, user comfort and convenience. In this study, feed-forward back-propagation artificial neural network (ANN) algorithm proposed based on acquired International Irregularity Index (IRI) data for highway structures, bridges culverts, obtained through laser profilometer measurements surface irregularity bituminous hot mix roads. Analysis ANN results were carried out training various hidden number networks output prediction, best estimation Results produced have been compared with experimental numerical extensive sets non-training data. As comparison study having average absolute mean relative errors as 12.68% 12.90% culverts provided very accurate results, model could be used obtain roads avoiding heavy duty collecting numerous field found more than models.

Language: Английский

---

Traffic-Forecasting Model with Spatio-Temporal Kernel

Electronics, Journal Year: 2025, Volume and Issue: 14(7), P. 1410 - 1410

Published: March 31, 2025

Within the realm of intelligent transportation systems, precise forecasting vehicular speed across individual road segments constitutes a fundamental task. This metric serves as pivotal indicator for evaluating extent network congestion and facilitating informed strategic planning. Contemporary methodologies predominantly employ recurrent neural networks (RNNs) to model temporal dependencies, while leveraging graph convolutional (GCNs) capture spatial dependencies within data. However, these methods fail integrate establish global dependencies. study introduces spatio-temporal kernel (STK-GCN), novel framework designed modeling traffic Specifically, we devise capable generating both matrices, which are subsequently utilized encoder–decoder architecture concurrently Furthermore, introduce convolution module enhance To demonstrate efficacy proposed STK-GCN, comprehensive experiments were carried out on two real-world datasets, namely METR-LA PEMS-BAY. The results indicate that our surpasses existing state-of-the-art methods.

Language: Английский

---

Neural Network Approach for Fatigue Crack Prediction in Asphalt Pavements Using Falling Weight Deflectometer Data

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3799 - 3799

Published: March 31, 2025

Fatigue cracking is a major issue in asphalt pavements, reducing their lifespan and increasing maintenance costs. This study develops an artificial neural network (ANN) model to predict the onset progression of fatigue cracking. The calibrated utilizing Falling Weight Deflectometer (FWD) testing data, alongside essential pavement characteristics such as layer thickness, air void percentage, binder proportion, traffic loads (Equivalent Single Axle Loads or ESALs), mean annual temperature. By analyzing these factors, ANN captures complex relationships influencing more effectively than traditional methods. A comprehensive dataset from Long-Term Pavement Performance (LTPP) program used for training validation. ANN’s ability adapt recognize patterns enhances its predictive accuracy, allowing reliable condition assessments. Model performance evaluated against real-world confirming effectiveness predicting with overall R2 0.9. study’s findings provide valuable insights rehabilitation planning, helping transportation agencies optimize repair schedules reduce research highlights growing role AI engineering, demonstrating how machine learning can improve infrastructure management. integrating ANN-based analytics, road enhance decision-making, leading durable cost-effective systems future.

Language: Английский

---

Current Status and Outlook of Roadbed Slope Stability Research: Study Based on Knowledge Mapping Bibliometric Network Analysis

Sustainability, Journal Year: 2025, Volume and Issue: 17(9), P. 4176 - 4176

Published: May 6, 2025

Landslide hazards on roadbed slopes pose significant safety risks, leading to casualties, property losses, and environmental damage. With the rapid expansion of global railway highway construction, slope stability has become a critical research focus. However, systematic reviews prospective studies based bibliometric analysis in this field remain limited; such lack is likely lead lag theoretical development field. To address gap, study analyzes 453 papers from 2014 2023 using Web Science (WOS) core collection tools like VOSviewer, CiteSpace, Bibliometrix R. This focuses following: (i) Visualizing trends through knowledge graphs, covering document quantity, authors, countries, keywords. (ii) The objectives, methods, specific objects, conditions literature are categorized discussed, limitations numerical simulation other shortcomings pointed out. (iii) Future directions, focusing actual working utilizing advanced flexible subroutine functions simulate complex with multi-physical coupling, discussed ensure accuracy sustainability road construction development. paper can help scholars comprehensively quickly understand status hotspots research, view providing support for future exploration.

Language: Английский

---

Predictive modeling of longitudinal cracking in CRCP using PSO-tuned gradient boosting machines

Journal of Engineering and Applied Science, Journal Year: 2025, Volume and Issue: 72(1)

Published: May 19, 2025

Abstract Longitudinal cracking poses a serious threat to the longevity and functionality of continuously reinforced concrete pavement (CRCP). Using structural, traffic, climatic data taken from Long-Term Pavement Performance (LTPP) database, this study presents machine learning system based on gradient boosting (GBM) optimized using particle swarm optimization (PSO) forecast longitudinal cracking. The proposed PSO-GBM model achieved lowest mean RMSE (2.661) highest R 2 (0.984) across fivefold cross-validation, outperforming baseline GBM, linear regression, random forest, artificial neural networks (ANN), support vector regression (SVR). Compared traditional untuned models, offers improved generalization stronger ability capture nonlinear interactions among variables. Feature importance sensitivity analyses identified L3 thickness, age, AADTT as key predictors. Despite model’s exceptional predictive accuracy, computational demands availability may limit its practical application. However, results offer useful information for transportation organizations looking improve maintenance planning techniques incorporate intelligent tools into management systems.

Language: Английский

---

Accelerometer-Based Pavement Classification for Vehicle Dynamics Analysis Using Neural Networks

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(21), P. 10027 - 10027

Published: Nov. 3, 2024

This research examines the influence of various pavement types on vehicle dynamics, specifically concentrating vertical acceleration and its implications for unsprung mass, including wheels suspension system. The objective this project was to categorize with accelerometer data, enabling a deeper comprehension impact road surface conditions stability, comfort, mechanical stress. Two categorization methods were utilized: neural network multinomial logistic regression model. Accelerometer data gathered while car navigated diverse terrain types, such as grates, potholes, cobblestones. model exhibited exceptional performance, 100% accuracy in categorizing all reached 97.14% accuracy. demonstrated efficacy differentiating intricate potholes surpassing which had difficulties these surfaces. These results underscore network’s effectiveness real-time surfaces, enhancing dynamics influenced by conditions. Future studies must tackle difficulty identifying analogous surfaces methodologies or integrating more attributes greater precision.

Language: Английский

---