New Technologies for Enhancing Energy Efficency in Buildings
Nuevas tecnologías para mejorar la eficiencia energética en los edificios
Acronym
NTech4Build
Key Words
Energy management, zero energy buildings, prediction models, people occupancy tracking systems, maintenance, machine learning.
Summary
In the European markets reference indices, the rise of some fossil fuels, such as oil, is pushing on the price of electricity in the markets both at the national level and in other countries of the European Union (EU). This increase in the cost at the continental level is due to various factors that do not have to be mutually exclusive:
- The fact that it has been possible to reach the production peak of the leading gas suppliers at the European level, Algeria and Russia.
- To various geopolitical tensions that affect supply, Algeria with Morocco or Russia with the European Union.
- Logistics and transportation problems due to the energy demand rebound once the pandemic caused by Covid-19 has been left behind.
These factors are accelerating the energy change from a model dependent on fossil fuels that are limited, polluting, and in the hands of a few countries, to another where renewable energies cover, if not all, a significant part of energy demand. In developed countries, the energy expenditure associated with residential or tertiary sector buildings is approximately one-third of the total energy consumed.
New technologies, as the Internet of Things (IoT), machine learning, Big Data, Augmented Reality (AR), and so on, can be used in energy management systems with the aim to reduce the energy consumption of the building and, thus, improve its efficiency. From the datasets acquired from these IoT devices is possible to develop through machine learning several models with this aim. Furthermore, the use of these techniques can help to:
- Detect anomalies that can reduce the energy wasted due to equipment malfunction and expensive faults.
- Characterize and model users behaviour inside the building that can be used to estimate where and when they develop their daily activities and adapt the HVAC systems working to this forecasting.
- Predict solar irradiation and photovoltaic panels production that drives to better management of the electricity produced by these panels and become the building to energy self-sufficient.
This project proposed the application of new digital technologies to reduce buildings energy consumption and, thus, drive their carbon footprint towards zero. Such a digital transition will be performed using IoT and machine learning algorithms and tested in an existing bioclimatic building located at the campus of the UAL, the CIESOL research center.
- Anomalies detection supported by AR. One passive measure for energy saving in buildings is the proper use of their electric appliances and equipment. To do so, anomalies detection of these items is mandatory to avoid malfunction and expensive faults. Through data-driven and knowledge based techniques is possible to check in real-time the operation of the main subsystems of the building. Besides, this objective will be supported with the use of AR in order to facilitate on-site checks and to reduce the time needed for maintenance tasks.
- Characterization and modeling of the users’ behaviour inside the building. Using an occupancy tracking strategy composed of anchors, tags and cameras is expected to create a system for counting people and estimating their trajectory inside the building. The collected data from this system will be the inputs to machine learning algorithms in order to extract trends and patterns and build models to estimate the users’ behaviour. Good estimations of occupancy counts can be helpful for building simulations and serve as the basis of control systems based on MPC.
- Predictions of the solar irradiation and the production of the photovoltaic panels. As is a current trend in modern buildings, where the use of renewable energies is mandatory for energy saving, the CIESOL building has a photovoltaic field to produce electricity. The production of the photovoltaic field depends on solar irradiation. Thus, two predictions will be made for an enhanced operation of the field: i) from data acquired by a pyranometer and using machine learning algorithms as ANNs. In this case, the solar irradiation will be forecasted with a prediction horizon of minutes and/or hours and, ii) from a video stream acquired by a camera focused on the photovoltaic panels. This will be used to estimate the electricity production depending on the amount of dust onto the photovoltaic panels.