Intelligent Architecture for Water Distribution: Robust Clustering to Demand Variations under Federated Learning

• Grant number: 24/08236-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: August 01, 2024
• End date: November 30, 2025
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: André Carlos Ponce de Leon Ferreira de Carvalho
• Grantee: Arthur Hiratsuka Rezende
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Company: Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC)
• Associated research grant: 20/09835-1 - IARA - Artificial Intelligence in the Remaking of Urban Environments, AP.PCPE
• Associated scholarship(s): 24/22560-2 - Establishing Reputations through Prototypes: Impartiality and Robustness in the Federated Context, BE.EP.IC

Abstract

UN projections indicate population growth alongside increasing urban concentration, factors that stress city infrastructure. Specifically regarding sanitation, Brazil experiences 37% losses in urban water distribution, highlighting the need to modernize this system to integrate it into the context of smart cities. Additionally, it is essential to consider the need to safeguard data privacy according to current legislation. Federated Learning allows for the aggregation of various systems in a smart city, locally or globally, resulting in truly integrated management while protecting sensitive data.It is assumed that the adoption of smart meters (for consumers) and sensors (in the distribution network) promotes greater knowledge about network conditions. In this scenario, the hypothesis is formulated that using generated data to group customers according to their consumption profile can reduce anomaly detection time (such as leaks and metering fraud) and optimize network pressure, reducing wear and tear. The main goal of the project is to verify the hypothesis both technically and economically, investigating the minimum proportion of consumers with smart meters to support it.By estimating the savings from efficient leak detection, it will be possible to calculate the Return on Investment (ROI) index to validate the use of smart sensors and meters, aiming for the project to assist operators and public managers in decision-making. The project's distinctive feature is comparing the impact of customer grouping through three paradigms: geographic criteria, optimized grouping through Information Theory, and customer grouping in the Federated paradigm, using methods with and without a priori knowledge of historical consumption data.