Description and objectives

Urban water distribution networks are complex and strategic infrastructures whose maintenance is of paramount importance, both to preserve the environment and prevent the waste of a valuable resource such as water, and to avoid significant service disruptions for end users.

Several research efforts have focused on defining data-driven or machine learning-based approaches for estimating degradation and enabling predictive maintenance of hydraulic structures, such as valves, pipelines, and various technological devices.

However, these approaches have generally been applied in very specific and limited contexts, thus losing generality. The AQUA-PREDICT research project aims to define a software architecture and propose a methodological approach to support and guide the design and development of generic applications in the context of predictive maintenance of urban water distribution networks.

The architecture will provide abstraction layers and components through which an application can be structured within an Edge/Cloud continuum and IoT sensor environment, leveraging artificial intelligence algorithms and tools for predictive functionalities.

The methodological approach will provide best practices for using the architecture. A case study will be presented to demonstrate the effectiveness of the approach using synthetic data.