Developing Machine Learning Models for Optimal Design of Water Distribution Networks Using Graph Theory-Based Features DOI Open Access

Iman Bahrami Chegeni,

Mohammad Mehdi Riyahi, Amin E. Bakhshipour

et al.

Water, Journal Year: 2025, Volume and Issue: 17(11), P. 1654 - 1654

Published: May 29, 2025

This study presents an innovative data-driven approach to optimally design water distribution networks (WDNs). The methodology comprises five key stages: Generation of 600 synthetic WDNs with diverse properties, optimized determine optimal component diameters; Extraction 80 topological and hydraulic features from the using graph theory; preprocessing preparing extracted established data science methods; Application six feature selection methods (Variance Threshold, k-best, chi-squared, Light Gradient-Boosting Machine, Permutation, Extreme Gradient Boosting) identify most relevant for describing Integration selected four machine learning models (Random Forest, Support Vector Bootstrap Aggregating, Machine), resulting in 24 ensemble models. Boosting-Light Machine (Xg-LGB) model emerged as choice, achieving R2, MAE, RMSE values 0.98, 0.017, 0.02, respectively. When applied a benchmark WDN, this accurately predicted diameters, 0.94, 0.054, 0.06, These results highlight developed model’s potential accurate efficient WDNs.

Language: Английский

Predicting Water Pipe Failures with Graph Neural Networks: Integrating Coupled Road and Pipeline Features DOI Open Access
Qunfang Hu, Yu Zhang, Wen Liu

et al.

Water, Journal Year: 2025, Volume and Issue: 17(9), P. 1307 - 1307

Published: April 27, 2025

The reliability of urban water distribution networks (WDNs) is critical for ensuring sustainable infrastructure management. However, traditional failure prediction models often overlook the complex interdependencies between pipelines and road networks, leading to suboptimal predictive accuracy. This study introduces a novel pipeline framework that leverages Graph Neural Networks (GNNs) incorporate coupled road–pipeline network features. By integrating traffic-related indicators, such as intersection proximity, pipeline–road angles, topology, this approach systematically assesses their impact on risk. A comparative evaluation various GNN architectures, including Convolutional (GCNs), Attention (GATs), GraphSAGE, demonstrates GraphSAGE achieves highest performance, significantly surpassing machine learning methods. findings underscore necessity incorporating topology into models, validating role spatial dependencies in accurately assessing risks. contributes advancing resilience modeling by providing robust supports proactive maintenance strategies enhances risk mitigation systems.

Language: Английский

Citations

0
An Optimal Probiotic Carrier: Multiple Steps Toward Selection and Application in Kombucha DOI Creative Commons

Tara Budimac,

Lato Pezo, Olja Šovljanski

et al.

Fermentation, Journal Year: 2025, Volume and Issue: 11(5), P. 256 - 256

Published: May 4, 2025

Kombucha is widely recognized as a functional beverage with potential probiotic effects, yet maintaining viability remains challenge due to the harsh conditions of fermentation. This study focuses on optimizing retention by identifying most effective carrier for Lactobacillus rhamnosus using multi-criteria decision-making approach. Five materials—pea protein, whey maltodextrin, inulin, and pectin—were assessed through three critical phases: evaluating encapsulated survival in different pH solutions, examining impact carriers kombucha fermentation, assessing stability during storage. The findings indicate that protein serves carrier, offering superior bacterial protection enhancing fermentation efficiency. Kinetic modeling further demonstrated significant correlation between survival, pH, titratable acidity, while artificial neural network models achieved high predictive accuracy (r2 > 0.9). Functional analysis revealed enriched encapsulates exhibited improved bioactivity, including enhanced antidiabetic properties α-glucosidase α-amylase inhibition, antihypertensive effects via ACE antihypercholesterolemic activity HMGCR inhibition. These suggest fortification contributes beverage’s overall health-promoting potential. Sensory evaluation highlighted slight modifications texture consumer acceptability remained high. underscores protein’s role an optimal significantly kombucha’s bio properties. results contribute advancements formulation, paving way development probiotic-enriched stability, appeal.

Language: Английский

Citations

0
Time-to-Failure Based Deterioration Factors of Water Networks: Systematic Review and Prioritization DOI
Beenish Bakhtawar,

Tarek Zayed,

Nehal Elshaboury

et al.

Reliability Engineering & System Safety, Journal Year: 2025, Volume and Issue: unknown, P. 111246 - 111246

Published: May 1, 2025

Language: Английский

Citations

0
Developing Machine Learning Models for Optimal Design of Water Distribution Networks Using Graph Theory-Based Features DOI Open Access

Iman Bahrami Chegeni,

Mohammad Mehdi Riyahi, Amin E. Bakhshipour

et al.

Water, Journal Year: 2025, Volume and Issue: 17(11), P. 1654 - 1654

Published: May 29, 2025

This study presents an innovative data-driven approach to optimally design water distribution networks (WDNs). The methodology comprises five key stages: Generation of 600 synthetic WDNs with diverse properties, optimized determine optimal component diameters; Extraction 80 topological and hydraulic features from the using graph theory; preprocessing preparing extracted established data science methods; Application six feature selection methods (Variance Threshold, k-best, chi-squared, Light Gradient-Boosting Machine, Permutation, Extreme Gradient Boosting) identify most relevant for describing Integration selected four machine learning models (Random Forest, Support Vector Bootstrap Aggregating, Machine), resulting in 24 ensemble models. Boosting-Light Machine (Xg-LGB) model emerged as choice, achieving R2, MAE, RMSE values 0.98, 0.017, 0.02, respectively. When applied a benchmark WDN, this accurately predicted diameters, 0.94, 0.054, 0.06, These results highlight developed model’s potential accurate efficient WDNs.

Language: Английский

Citations

0