282. Data-Driven Prediction of Sewer Blockages: A Spatiotemporal Analysis and Machine Learning Approach
Invited abstract in session WC-17: Combinatorial Optimization and Data Processing, stream Combinatorial Optimization.
Wednesday, 12:30-14:00Room: Esther Simpson 2.08
Authors (first author is the speaker)
| 1. | Lorena Pradenas
|
| Ingeniería Industrial, Universidad de Concepción | |
| 2. | Iván Veloso
|
| Departamento de Ingeniería Industrial, Universidad de Concepción | |
| 3. | Mauricio Vega-Hidalgo
|
| Industrial Engineering, University of Concepción | |
| 4. | Víctor Parada
|
| Universidad de Santiago de Chile |
Abstract
This study researches the occurrence of blockages in urban sewer networks due to the accumulation of fats, oils, and grease (FOG), along with solid waste. A data-driven methodology was employed to conduct a spatiotemporal analysis of blockage events and develop predictive binary classification models. The study area was represented as a directed graph, integrating hydraulic and infrastructural attributes. Three predictive models were implemented: logistic regression, random forests, and neural networks, each evaluated using standard classification metrics. Logistic regression demonstrated an optimal trade-off between sensitivity (84%) and practical applicability, whereas more complex models exhibited superior performance in capturing non-linear patterns. The findings underscore the significance of data-driven strategies in enhancing preventive maintenance protocols, particularly in high-risk zones, thereby improving the resilience and efficiency of sewer management systems.
Keywords
- Machine Learning
- Water Management
- Environmental Management
Status: accepted
Back to the list of papers