Swegon brings predictive maintenance to HVAC services with Statwolf’s AI
An industrial AI solution to turn data from connected HVAC units into anomaly indicators, alerts and tools supporting service activities, integrated into the Swegon digital experience.
“The value of working with Statwolf was bringing AI into an existing ecosystem without disrupting the operational platforms. The project helped us make data, pipelines and integrations more robust, turning the model into a service that is genuinely usable in production.”
“Predictive maintenance applied to HVAC machines requires close integration between domain expertise and Artificial Intelligence technologies. Working with Statwolf allowed us to turn this experience into a concrete digital service, able to support service activities and offer more value to our customers.”
Context
Swegon, through its BlueBox brand, is one of the leading names in heat pumps, chillers and HVAC solutions. The HVAC market in which Swegon operates is growing steadily, driven in particular, for heat pumps, by the energy transition and by the spread of increasingly advanced technologies. The evolution of systems towards more complex architectures, the introduction of new refrigerants and the continuous updating of regulations require ever higher levels of specialisation, while the availability of experienced maintenance technicians is decreasing. To face this challenge, Swegon decided to invest in a proprietary cloud architecture, “INSIDE”, and in digital solutions able to support technical staff, make the most of the available know-how and improve the effectiveness of field service activities.
In this context Swegon decided to further strengthen its digital services platform by developing a predictive maintenance service for its hydronic units, heat pumps and chillers. Swegon worked with Statwolf, which provided specialist expertise in industrial Artificial Intelligence and data valorisation, in order to develop “BlueBox Predictive Maintenance”, a digital service designed to identify anomalous behaviour of HVAC units early, supporting proactive maintenance activities and helping to reduce downtime.
Through continuous analysis of operating data, the system generates anomaly indicators and provides service staff with information useful to support diagnosis and the planning of interventions.
The challenge
Swegon wanted to go beyond traditional monitoring of HVAC units, introducing a concrete and operational predictive maintenance service.
The goal was not the simple collection of machine data, but the creation of a solution able to:
- identify anomalous behaviour promptly;
- generate reliable alerts;
- support diagnosis and root cause analysis;
- be scalable both in terms of number of units in the field and of variability of unit types
- integrate within the Swegon cloud platform
The main challenge was turning large amounts of data from HVAC units into reliable, interpretable information usable by service staff, while maintaining high standards of robustness and scalability.
The approach
Statwolf and Swegon developed the project through continuous collaboration between Artificial Intelligence expertise and applied knowledge of the HVAC domain.
The solution implements data acquisition and organisation mechanisms that allow correct integration with the know-how developed by Swegon’s experts, making the data usable by the dedicated Artificial Intelligence models.
The effectiveness of the solution comes from the close integration between technological expertise and knowledge of the real behaviour of the machines, producing results that are reliable and easy to interpret in the operational context.
The project was developed with a strongly collaborative approach: Statwolf contributed expertise in industrial AI and data platforms, while Swegon brought its experience in the design of HVAC machines, their operating logic and service activities.
The solution
The technology developed by Statwolf is the AI engine behind the BlueBox Predictive Maintenance service.
It can be described through six main components.
Connected HVAC assets
The connected HVAC units are the information source of the system, providing the data needed for continuous performance monitoring.
Data foundation
Acquisition, integration and organisation of the data coming from the connected units, in order to make it available to the analysis processes.
Domain Knowledge Integration
The applied knowledge built up by Swegon’s experts is integrated into the analysis process to improve the quality and reliability of the results.
Statwolf AI Analysis
Data analysis makes it possible to generate easily interpretable anomaly indicators, useful to highlight any deviations from the expected behaviour of the units.
Operational Reliability
The solution includes dedicated tools for continuous monitoring of operating performance and for maintaining the reliability of the service over time.
Digital Service Integration
The results are integrated into Swegon’s digital services, enabling the visualisation of anomaly indicators, the generation of notifications and support to troubleshooting activities.
Why Statwolf
The Swegon case study represents Statwolf’s mission well: not just the development of algorithms, but the ability to bring industrial AI into production. Through its platform, Statwolf represents an enabler for production-grade AI solutions, with data integration, interoperability, MLOps, data governance and cloud, on-premises or hybrid deployment.
In the Swegon case, this translates into a solution that combines:
- integration with existing systems;
- interpretable AI;
- domain knowledge;
- robust pipelines;
- MLOps;
- release onto a platform used by the end customer.
The goal of the project was not simply to develop an Artificial Intelligence algorithm, but to turn data from HVAC units into a reliable, integrated digital service that is genuinely usable in service activities. The collaboration between Swegon and Statwolf made it possible to combine technological expertise and applied knowledge, delivering a scalable solution able to support proactive maintenance of the installed systems.
The service dashboards
