Edge AI-powered digital twins target OpEx cuts in smart buildings
Edge AI-powered digital twins are being explored as a way to rein in the “phantom load” that quietly drives up electricity use in commercial buildings. Researchers at the University of Glasgow’s James Watt School of Engineering have built and tested a prototype that links smart plugs, environmental sensors and local AI decision-making to switch idle devices off more intelligently. Early results suggest that this kind of architecture could translate into meaningful operational expenditure savings for facilities teams.
For eeNews Europe readers, the work points to how edge AI-powered digital twins might influence next-generation building automation, from the choice of IoT connectivity and edge compute to how energy dashboards feed into corporate sustainability targets. It also touches on an emerging intersection between embedded AI, power electronics and facilities IT that could shape requirements for future design-in projects.
Tackling phantom load with edge intelligence
In office and campus environments, phantom load from devices left in standby can account for around a third of total electricity usage, yet it is often treated as background noise in the energy budget. The Glasgow team’s prototype instead links plug-level smart meters and environmental sensors via LoRaWAN to an edge server that hosts the “digital twin” of each monitored asset.
A local AI layer then replaces simple timer-based switching with a fuzzy-logic framework that considers several metrics, including a user habit score, device activity score and confidence score. Rather than cutting power after a fixed interval, the system can maintain the current state, delay a decision, shut down a device or prompt the user to confirm that a background workload is still running. The aim is to curb idle consumption while avoiding disruptions that would cause users to bypass automation altogether.
Dr Ahmad Taha, Lecturer for Autonomous Systems & Connectivity at the James Watt School of Engineering, who is leading the work, said: “I’m a firm believer in the idea that that small, collective actions on climate issues can have big effects, and phantom power use is an obvious candidate for that kind of action.”
From lab results to building-scale savings
To validate the concept, the researchers deployed their edge AI-powered digital twins system in a university laboratory using smart plugs and LoRaWAN-connected sensors. In that environment, weekly power consumption per monitored workstation was reduced by roughly 40 percent, with phantom loads themselves cut by up to 82 percent. When extrapolated to a 500-device deployment using current UK electricity price caps, the model suggests potential annual savings above £9,000.
The implementation uses a containerised stack that will be familiar to many embedded and IoT developers: Docker-hosted services including an MQTT broker for messaging, Node-RED for data handling and InfluxDB for time-series storage. A forecasting module based on an LSTM (Long Short-Term Memory) model trains on short histories of consumption data to predict the next day’s demand profile, giving facilities teams a way to anticipate peaks rather than react after the fact.
User acceptance remains a key constraint, so the architecture incorporates an “anti-oscillation” filter to avoid rapid on–off cycling that could annoy staff or stress hardware. Alongside energy savings, the team also points to a possible impact on asset lifetimes. “Secondly, by reducing devices’ use of electricity, it could help reduce the need to replace older devices with newer, more power-efficient ones. That in turn could help organisations save on equipment costs in an increasingly challenging economic environment.”
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
