Partial discharge detection
Ultrasonic-capable microphones at panels and bays, edge detectors, severity trends to the maintenance system.
Industry
Power, utilities and heavy electrical
Partial discharge and rotating-machine monitoring for plant substations, captive power and generation.
The situation
What plants in this sector tell us
- Insulation faults that announce themselves for months between annual survey rounds.
- Handheld ultrasonic surveys that only see the asset on the day the technician walks past it.
- Outage windows too short and too rare to install anything invasive.
- Severity judged by an experienced pair of ears, and lost when that person retires.
Insulation fails slowly and audibly, and your maintenance team already knows it: handheld ultrasonic and TEV surveys of switchgear are standard practice. What is missing is the version that never goes home. Our published work on partial discharge detection with airborne acoustics and deep neural networks makes that survey continuous and unattended, on switchgear, bushings, insulators and cable terminations: no outage for installation, no electrical connection to the asset, and no raw data leaving the site. Oil-filled transformers are monitored with tank-mounted sensors instead. Relevant to every steel, cement and chemical plant running its own high-voltage distribution, to captive power plants and independent producers, and to utilities.
Where we help
Where our methods apply
Rotating electrical machines
Motor, generator and turbine condition from vibration and acoustics, fused with drive or SCADA data where available, alongside the monitoring system you already run.
Transformers and reactors
Tank-mounted acoustic-emission sensors for internal partial discharge, and vibro-acoustic signatures of tap-changer operation.
Multi-site fleets
Federated learning across substations without centralising recordings.
EvidencePartial discharge monitoring with deep neural networks (INTER-NOISE 2021), research led by our founder; project experience with turbines, motors, generators and gearboxes in German industry.
By asset
The machines we listen to in this sector
Bearings and gearboxes
Wear, lubrication loss and race defects from vibration and airborne acoustics; on-node classification for fleets.
Motors, generators and e-drives
In service: bearing, rotor and mechanical faults from vibration and sound, fused with drive or PLC data. At the end of the line: acoustic pass/fail testing of new motors, e-axles, gearboxes and compressors with learned limits instead of fixed thresholds.
Turbines, fans and compressors
Load, imbalance and abnormal operation from acoustic and vibration signatures.
Switchgear, bushings and terminations
Partial discharge at switchgear, bushings and terminations detected without contact; oil-filled transformer tanks and tap-changers with contact sensors instead.
The whole site
Acoustic surveillance on edge devices: impacts, leaks, arcing and unusual events flagged to the maintenance system. Offered as an add-on to a monitoring project, not as a first step.
How we would start
The services this sector uses most
Almost every engagement begins with a feasibility study or a discovery call, because the cheapest thing to establish first is whether the signature is there at all.
Multi-modal sensing + edge AIRuggedised sensor units and on-device AI.Open service page Co-development R&D and PoCApplied R&D on your process, with your engineers.Open service page Expert consultation & advisorySenior engineering time, by the day or on retainer.Open service page Sovereign AI & federated learningOn-premise AI and federated learning across plants.Open service page
Start here
Discuss a project in Power & heavy electrical
Tell us the asset or the station, what goes wrong and how often. We answer every enquiry within 48 hours on working days.
- Your data stays on site; on-premise and on-device deployment is the default
- We work alongside the monitoring system you already run
- Honest go / no-go before any large spend
Ready to find out if it works on your machines?
Start with a free 30 to 45 minute discovery call. Bring the problem, the recordings you already have and your questions. We will tell you honestly whether sensing-based AI is the right tool and what the next step would cost in days, not months.