M/F) PhD Position: Multi-Source Data Fusion for the Modeling, Data Enrichment, and Causal Analysis of Urban Water Networks

Descrizione

M/F) PhD Position: Multi-Source Data Fusion for the Modeling, Data Enrichment, and Causal Analysis of Urban Water Networks Sintesi in italiano (traduzione automatica): L'organizzazione offre una posizione di dottorato nell'ambito del progetto PRCI LUCAS, focalizzato sulla fusione di dati multi-sorgente per la modellazione e l'analisi delle reti idriche urbane. La ricerca si svolgerà presso il Centre de Recherche en Informatique de Lens (CRIL) e il laboratorio IUSTI. Il dottorando, supervisionato da Salem Benferhat e Carole Delenne, si occuperà di estrazione automatica di informazioni da dati non strutturati, sviluppo di metodi di fusione dei dati e rilevamento automatico di anomalie. È richiesta una laurea in ingegneria o in un campo affine, con competenze in GIS e analisi dei dati. Il lavoro prevede l'uso di dati reali per migliorare la gestione delle infrastrutture idriche e la resilienza agli eventi estremi, come le inondazioni urbane. The PhD thesis will be funded within the framework of the PRCI project entitled LUCAS (Leverage External Data for Enhanced Understanding and Causal Attribution of Anomalies in Water Network Systems). The research activities will be carried out at the Centre de Recherche en Informatique de Lens (CRIL) and within the IUSTI laboratory (“Institut Universitaire des Systèmes Thermiques Industriels”). The PhD student will be supervised by the two thesis advisors, Salem Benferhat (CRIL) and Carole Delenne (IUSTI). Urban water networks are critical infrastructures whose management relies on a wide variety of data sources: Geographic Information Systems (GIS), Closed-Circuit Television (CCTV) inspection videos, technical PDF reports, mapping data, and asset management databases, produced by different stakeholders and for various purposes. These data are often heterogeneous, incomplete, inconsistent, or affected by uncertainties, limiting their use for asset knowledge, infrastructure maintenance, and urban risk management. In this context, the central research question of this PhD thesis is the following: How can these imperfect data sources be jointly exploited to build a unified, consistent, and enriched representation of urban water networks, while enabling anomaly identification, causal analysis, and improved infrastructure resilience in the face of extreme events, particularly urban flooding? To address this challenge, the thesis will pursue several scientific objectives: automating information extraction from unstructured data, proposing a common representation of multi-source data, developing data fusion and knowledge enrichment methods, automatically detecting anomalies, and implementing causal attribution mechanisms. The first stage will focus on extracting and structuring information from unstructured sources such as CCTV inspection videos and inspection reports. These documents contain extensive information about pipeline conditions, observed defects, hydraulic characteristics, and network operating conditions. A matching process with asset management and mapping data will make it possible to link these observations to the physical components of the network despite uncertainties and missing data. The second stage will aim to integrate all available information into a common knowledge representation language. The thesis will particularly investigate approaches based on attributed graphs to model infrastructures, their topological relationships, inspection-derived observations, and the levels of uncertainty associated with the data. This representation will provide a foundation for fusion and enrichment mechanisms that exploit the complementarity of different sources in order to progressively improve the completeness and consistency of available knowledge. The third stage will focus on multi-source automatic anomaly detection and the analysis of their potential causes. The objective will not only be to identify existing defects or malfunctions in networks, but also to better understand the factors likely to cause them. Particular attention will be given to integrating contextual information related to the urban environment, infrastructures. All developed methods will be validated using real-world datasets from the Montpellier metropolitan area, by assessing their ability to enrich existing GIS databases, reduce inconsistencies between data sources, and improve knowledge of underground water networks. Annuncio in inglese. Fonte: Euraxess (Commissione europea).