III.3 Related work 143 Graphical Neural Networks and DRL to model the sewer network structure, focusing on grouping maintenance actions by leveraging sewer main proximity. Machine Learning Machine Learning (ML) enables computers to improve their performance on specific tasks by learning from data, with extensive applications in various domains, including sewer asset management. A review on this is provided by Marc Ribalta and Rubión, 2023. In the realm of condition assessment, techniques such as Knearest neighbours, Support Vector Machines (SVM), Random Forest, Principal Component Analysis, and Gate Recurrent Unit are utilised for concrete defect detection and classification, as detailed by Gueye, Y. Wang, and Mushtaq, 2023, while Cheng and M. Wang, 2018 and Fang, Guo, Q. Li, et al., 2020 implement Deep Learning and methods like Isolate Forest, One-Class SVM, Local Outlier Factors, and Gaussian Distributed-based approaches for video sequences anomaly detection. For degradation modelling of sewer mains, Random Survival Forest, SVM, and Random Forest are applied to model sewer degradation and determine the distribution of time-to-failure, as explored by Caradot, Riechel, Fesneau, et al., 2018, Laakso, Kokkonen, Mellin, et al., 2019, and Hernández, Caradot, Sonnenberg, et al., 2021. Additionally, ML aids in maintenance decision-making by using data from sewer overflows and Decision Trees, as demonstrated by Montserrat, Bosch, Kiser, et al., 2015. Decision-support frameworks Decision-support tools, as high-level and generic management approaches, enhance decision-making by integrating various concepts, including those previously discussed. DeSilva, Burn, Tjandraatmadja, et al., 2005 explores sewer main leakage and introduces a decision support tool for rehabilitation prioritisation, utilising soil models, pipe properties, and operational conditions. Similarly, Ana and Bauwens, 2007 o!ers a decision support tool that employs multiple sewer management and rehabilitation models. Breysse, Vasconcelos, and Schoefs, 2007 combines social and technical cost indicators to provide a comprehensive tool for managers to assess alternatives. “Urban stormwater drainage management: The development of a multicriteria decision aid approach for best management practices” 2007 assists decision-makers in ranking options using multi-criteria analysis. Arthur, Crow, Pedezert, et al., 2009 suggests a holistic approach based on Failure Mode, E"ects & Criticality Analysis (FMECA), implementable with limited information without additional data collection. M.A. Cardoso and Silva, 2016 introduces the AWARE-P procedure, which combines various decision-support tools and methods considering performance, costs, and risk over an analysis horizon. Obradovi$, %perac, and Marenjak, 2019 discusses the use of expert systems to support sewer maintenance optimisation. Lin, Yuan, and Tovilla, 2019 and Caradot, Sampaio, Guilbert, et al., 2021 focus on an integrated approach that considers modelling long-term sewer
RkJQdWJsaXNoZXIy MjY0ODMw