Pressure and flow estimation in water distribution networks (WDNs) allows water management companies to optimize their control operations. For many years, mathematical simulation tools have been the most common approach to reconstructing an estimate of the WDNs hydraulics. However, pure physics‐based simulations involve several challenges, for example, partially observable data, high uncertainty, and extensive manual calibration. Thus, data‐driven approaches have gained traction to overcome such limitations. In this work, we combine physics‐based modeling and graph neural networks (GNN), a data‐driven approach, to address the pressure estimation problem. Our work has two main contributions. First, a training strategy that relies on random sensor placement making our GNN‐based estimation model robust to unexpected sensor location changes. Second, a realistic evaluation protocol that considers real temporal patterns and noise injection to mimic the uncertainties intrinsic to real‐world scenarios. As a result, a new state‐of‐the‐art model, GAT with Residual Connections, for pressure estimation is available. Our model surpasses the performance of previous studies on several WDNs benchmarks, showing a reduction of absolute error of ≈40% on average.

In addition to providing for a more reliable distribution infrastructure, the smart grid promises to give the end users better pricing and usage information. It is thus interesting for them to be ready to take advantage of features such as dynamic energy pricing and real-time choice of operators. In this work, we propose a system to monitor and control an office environment and to couple it with the smart grid. The idea is to schedule the operation of devices according to policies defined by the users, in order to minimize the cost of operation while leaving unaffected user comfort and productivity. The implementation of the system and its testing in a living lab environment show interesting economic savings of an average of about 35% and in some cases even overall energy savings in the order of 10% for a building equipped with renewable generation plants, and economic and energy savings of 20% and 10%, respectively, for a building without local renewable installations.

By virtue of the steady societal shift to the use of smart technologies built on the increasingly popular smart grid framework, we have noticed an increase in the need to analyze household electricity consumption at the individual level. In order to work efficiently, these technologies rely on load forecasting to optimize operations that are related to energy consumption (such as household appliance scheduling). This paper proposes a novel load forecasting method that utilizes a clustering step prior to the forecasting step to group together days that exhibit similar energy consumption patterns. Following that, we attempt to classify new days into pre-generated clusters by making use of the available context information (day of the week, month, predicted weather). Finally, using available historical data (with regard to energy consumption) alongside meteorological and temporal variables, we train a CNN-LSTM model on a per-cluster basis that specializes in forecasting based on the energy profiles present within each cluster. This method leads to improvements in forecasting performance (upwards of a 10% increase in mean absolute percentage error scores) and provides us with the added benefit of being able to easily highlight and extract information that allows us to identify which external variables have an effect on the energy consumption of any individual household.

Privacy restrictions hinder the sharing of real-world Water Distribution Network (WDN) models, limiting the application of emerging data-driven machine learning, which typically requires extensive observations. To address this challenge, we propose the dataset DiTEC-WDN that comprises 36,000 unique scenarios simulated over either short-term (24 hours) or long-term (1 year) periods. We constructed this dataset using an automated pipeline that optimizes crucial parameters (e.g., pressure, flow rate, and demand patterns), facilitates large-scale simulations, and records discrete, synthetic but hydraulically realistic states under standard conditions via rule validation and post-hoc analysis. With a total of 228 million generated graph-based states, DiTEC-WDN can support a variety of machine-learning tasks, including graph-level, node-level, and link-level regression, as well as time-series forecasting. This contribution, released under a public license, encourages open scientific research in the critical water sector, eliminates the risk of exposing sensitive data, and fulfills the need for a large-scale water distribution network benchmark for study comparisons and scenario analysis.

The advances of pervasive technology offer new standards for user comfort by adding intelligence to ubiquitous home and office appliances. With intelligence being the core of some newly constructed buildings, it is important to design a scalable, robust, context-aware architecture, which not only has enough longevity and evolving capabilities to sustain itself over the building's lifetime, but also provides enough potential for additional features to be added to the core Building Management Systems (BMS). Such features may include energy preservation system, or activity-recognition techniques. Service-Oriented Architecture (SOA) principles provide great tools that can be applied to the smart buildings design, however certain specifics of pervasive systems should be taken into account, such as high heterogeneity of available devices and capabilities. In this paper we propose an architecture for smart pervasive applications, which is based on SOA principles and is specifically designed for long-term applicability, scalability, and evolution capabilities of a BMS. We validate our proposal by implementing a smart office on the premises of the Technical University of Eindhoven and showing that it complies with the requirements of scalability and robustness, at the same time being a viable BMS.