Szczegóły publikacji
Opis bibliograficzny
Machine learning in predicting failures of buried water supply networks affected by mining impacts / Leszek Chomacki, Janusz RUSEK, Leszek Słowik // Scientific Reports [Dokument elektroniczny]. — Czasopismo elektroniczne ; ISSN 2045-2322 . — 2026 — vol. 16 iss. 1 art. no. 8465, s. 1–11. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 9–10, Abstr. — Publikacja dostępna online od: 2026-02-12
Autorzy (3)
- Chomacki Leszek
- AGHRusek Janusz
- Słowik Leszek
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 166966 |
|---|---|
| Data dodania do BaDAP | 2026-04-16 |
| Tekst źródłowy | URL |
| DOI | 10.1038/s41598-026-39874-w |
| Rok publikacji | 2026 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Creative Commons | |
| Czasopismo/seria | Scientific Reports |
Abstract
The article presents the research results concerning the failure predicting of buried water supply pipelines located in mining areas. Such underground infrastructure components are common in mining areas and generate many difficult problems related to their proper use. Network failures are difficult to predict and prevent, resulting in decreased comfort for residents and increased water consumption, to the detriment of the natural environment. From this perspective, it is a broader issue related to sustainable development. To address this problem, machine learning methods have been proposed as effective tools for modeling multifactorial and complex phenomena with implicit behavior. The study implemented five different approaches to describe the phenomenon under investigation. Each of them was based on the use of ML methods previously selected for this purpose. The results indicate that the most effective methods for modeling the failure process of underground water supply networks are eXtreme Gradient Boosting (RMSE 0.43 for training set and 2.03 for test set) and Support Vector Machine (RMSE 0.66 and 0.71). During the research, factors determining the process of underground water pipeline failures were identified and graded according to their relevance. It turned out that among the input factors, the most important is the length of the analyzed section of the network, followed by its technical age. Horizontal strain and diameter proved to be less important than the two main factors mentioned above. It has been demonstrated that the ML methods used can be very helpful in monitoring and managing the failures of underground water pipelines affected by mining impacts. The conducted research may serve as an inspiration for further research on the failure rate of various infrastructure networks in mining areas.
