Explainable ensemble models for predicting wall thickness loss of water pipes
Water Distribution Networks (WDNs) are susceptible to pipe failures with significant consequences. Predicting wall-thickness loss in pipes is vital for proactive maintenance and asset management. This study develops optimized, explainable machine learning models for this purpose. Data from four WDNs...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2024-04-01
|
Series: | Ain Shams Engineering Journal |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2090447924000054 |
_version_ | 1797238060506677248 |
---|---|
author | Ridwan Taiwo Abdul-Mugis Yussif Mohamed El Amine Ben Seghier Tarek Zayed |
author_facet | Ridwan Taiwo Abdul-Mugis Yussif Mohamed El Amine Ben Seghier Tarek Zayed |
author_sort | Ridwan Taiwo |
collection | DOAJ |
description | Water Distribution Networks (WDNs) are susceptible to pipe failures with significant consequences. Predicting wall-thickness loss in pipes is vital for proactive maintenance and asset management. This study develops optimized, explainable machine learning models for this purpose. Data from four WDNs located in Canada and the USA are collected and preprocessed. Decision Tree, Random Forest (RF), XGBoost, LightGBM, and CatBoost are employed, with optimized hyperparameters via Tree-Structured Parzen Estimator. The proposed framework performance is assessed using dissimilarity-based and similarity-based metrics. Hyperparameter optimization substantially enhances predictive performance such that the mean absolute error of RF improved by 20.51%. Based on the evaluation metrics, the Copeland algorithm was employed to rank the models, and CatBoost emerged as the best-performing model with a Copeland score of 4, followed by XGBoost and RF. The Taylor Diagram offers a visual representation of the linear proportionality between observed and predicted values across various models, with CatBoost and XGBoost showing strong alignment. SHAP analysis identifies age, diameter, and length as key contributors. The optimized models proactively identify potential pipe failures, enhancing maintenance and WDN management. |
first_indexed | 2024-04-24T17:29:38Z |
format | Article |
id | doaj.art-ad2dc1ba97a34b3d90a1ea0701e0ae5d |
institution | Directory Open Access Journal |
issn | 2090-4479 |
language | English |
last_indexed | 2024-04-24T17:29:38Z |
publishDate | 2024-04-01 |
publisher | Elsevier |
record_format | Article |
series | Ain Shams Engineering Journal |
spelling | doaj.art-ad2dc1ba97a34b3d90a1ea0701e0ae5d2024-03-28T06:37:34ZengElsevierAin Shams Engineering Journal2090-44792024-04-01154102630Explainable ensemble models for predicting wall thickness loss of water pipesRidwan Taiwo0Abdul-Mugis Yussif1Mohamed El Amine Ben Seghier2Tarek Zayed3Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hum, Hong KongDepartment of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hum, Hong KongDepartment of Built Environment, Oslo Metropolitan (OsloMet) University, Oslo, Norway; Corresponding author.Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hum, Hong KongWater Distribution Networks (WDNs) are susceptible to pipe failures with significant consequences. Predicting wall-thickness loss in pipes is vital for proactive maintenance and asset management. This study develops optimized, explainable machine learning models for this purpose. Data from four WDNs located in Canada and the USA are collected and preprocessed. Decision Tree, Random Forest (RF), XGBoost, LightGBM, and CatBoost are employed, with optimized hyperparameters via Tree-Structured Parzen Estimator. The proposed framework performance is assessed using dissimilarity-based and similarity-based metrics. Hyperparameter optimization substantially enhances predictive performance such that the mean absolute error of RF improved by 20.51%. Based on the evaluation metrics, the Copeland algorithm was employed to rank the models, and CatBoost emerged as the best-performing model with a Copeland score of 4, followed by XGBoost and RF. The Taylor Diagram offers a visual representation of the linear proportionality between observed and predicted values across various models, with CatBoost and XGBoost showing strong alignment. SHAP analysis identifies age, diameter, and length as key contributors. The optimized models proactively identify potential pipe failures, enhancing maintenance and WDN management.http://www.sciencedirect.com/science/article/pii/S2090447924000054Water pipelinesWall thicknessMachine learningEnsemble learningSHAP |
spellingShingle | Ridwan Taiwo Abdul-Mugis Yussif Mohamed El Amine Ben Seghier Tarek Zayed Explainable ensemble models for predicting wall thickness loss of water pipes Ain Shams Engineering Journal Water pipelines Wall thickness Machine learning Ensemble learning SHAP |
title | Explainable ensemble models for predicting wall thickness loss of water pipes |
title_full | Explainable ensemble models for predicting wall thickness loss of water pipes |
title_fullStr | Explainable ensemble models for predicting wall thickness loss of water pipes |
title_full_unstemmed | Explainable ensemble models for predicting wall thickness loss of water pipes |
title_short | Explainable ensemble models for predicting wall thickness loss of water pipes |
title_sort | explainable ensemble models for predicting wall thickness loss of water pipes |
topic | Water pipelines Wall thickness Machine learning Ensemble learning SHAP |
url | http://www.sciencedirect.com/science/article/pii/S2090447924000054 |
work_keys_str_mv | AT ridwantaiwo explainableensemblemodelsforpredictingwallthicknesslossofwaterpipes AT abdulmugisyussif explainableensemblemodelsforpredictingwallthicknesslossofwaterpipes AT mohamedelaminebenseghier explainableensemblemodelsforpredictingwallthicknesslossofwaterpipes AT tarekzayed explainableensemblemodelsforpredictingwallthicknesslossofwaterpipes |