Predictive fault analytics for power networks
Predict electrical faults weeks before they cause outages
IntelliView® reads the disturbance recordings your digital substations already produce — and tells you which asset is failing, why, and when.
In service with utilities and industry
20+ deployments, pilots and projects across Europe, North America, the Middle East and Asia.
15 days
average warning before the fault
time to schedule an inspection, not answer a callout
100+
outages predicted in advance
documented across customer deployments
1M+
disturbance recordings analyzed
every one screened, classified and scored
0
proprietary sensors
runs on standard substation measurements
The pressure
You cannot inspect everything
Four forces are arriving at the same assets at the same time: estates built decades ago are reaching the end of their designed life, electrification pushes more current through the same plant, distributed renewables make flows volatile — and under incentive regulation, interruptions feed straight into revenue. Crews and engineering hours are the scarcest resource you have. The only lever left is knowing which asset to look at first — and that is a data problem, not a manpower problem.
The shift
The warning signs are already in your data
Every abnormal event triggers a high-resolution waveform capture — holding the early signatures of insulation decay, loose connectors and intermittent earth contact. Today those files are opened only after an outage, for forensics. IntelliView® reads them continuously — and tells you when the pattern starts to resemble a failure.
Evolving fault signatures are flagged weeks before a breaker ever trips.
The platform
— Which asset is failing, why, and when
Risk you can act on
Rolling fault risk for every substation, bay and feeder — over one-day, one-week and ten-day horizons. The stream of recordings condenses into one ranked answer: which bays deserve attention this week, and why.
Explainable, not a black box
Physics-informed AI on 400 expert-designed waveform features. Every warning carries its predicted root cause and likely source — the engineer decides, not the model.
Runs on the data you already have
Standard-compliant relays, fault recorders and power-quality meters already write COMTRADE records or IEC 61850 sampled values — IntelliView® reads them as they are, with no proprietary hardware of its own. Where a substation isn't digitalized yet, our partners ABB and EcoPhi supply the measurement layer — along with the wider benefits of substation digitalization.
“A system that can predict problems and identify causes could be invaluable in maintaining the resilience of the transmission system — not only for NYPA but other utilities as well.”
Alan Ettlinger — Senior Director of Research, New York Power Authority “We can now not only locate faults but also predict them. We want to work proactively to prevent costly failures.”
Lukas Lukiewicz — Grid Analyst, Härryda Energi “The portfolio synergies between ABB and Eneryield will help create one of the most advanced prediction systems for power networks.”
Sami Raitakoski — Global Product Group Manager Grid Components, ABB “Eneryield is a great example of how deep tech solutions can successfully spin out from research.”
Björn Westman — Investment Director, Chalmers Ventures
Proof
Proven in the field since October 2020
Distribution and transmission operators in Europe, North America and the Middle East — measured on real operational customer data.
Finland — distribution
Vaasan Sähköverkko
On two distribution feeders at the Petalax substation, IntelliView® has been in continuous service since October 2020 — predicting 64 faults in advance, with warnings up to three weeks before the interruption.
64
faults predicted in advance
15 days
average warning
Oct 2020
in service since
United States — transmission
New York Power Authority
Cable, reactor and transformer banks monitored since October 2021 — nine faults predicted in advance, with warnings up to three weeks ahead, using the sensors NYPA already had.
9
faults predicted in advance
15 days
average warning
Oct 2021
in service since
Saudi Arabia — utility
Saudi Electricity Company
Three substations and 37 bays at 33 kV and 69/13.8 kV, live since March 2026 with our partner EcoPhi — five faults already predicted in advance, with warnings up to two weeks ahead.
5
faults predicted in advance
15 days
average warning
Mar 2026
in service since
The first step
Find out in six weeks, on your own data
Send us historical disturbance recordings and your own fault log. We replay the models across that window and report what we would have flagged, when, and why — measured against criteria you set before we start.
Request a replayScoped fee
Agreed up front, sized to the data scope.
Nothing installed
No integration, no substation visit.
You set the bar
Success metric agreed before we run.
Six weeks
From receiving the data to the result.
Backed by
A research company that ships product
Eneryield is a Chalmers spin-out, founded 2019 and in continuous customer operation since 2020 — built on more than a decade of the founders' research in machine learning and power systems: physics-informed, peer-reviewed, patented. Backed by ABB and Chalmers Ventures.
- ABB Draknästet winner 2019
- EPRI Incubatenergy Labs 2021 cohort
- Svenska Energipriset 2025 — finalist
Latest
News & insights
Meet Eneryield at CIGRE Paris Session 2026
We're exhibiting with our partner EcoPhi at the world's largest power-systems conference, August 23–28 in Paris — book a meeting in advance.
Tidningen Energi: AI is keeping the grid in Härryda running
Tidningen Energi features our live deployment with Härryda Energi — AI fault prediction making a Swedish distribution grid more resilient.
Preventing Substation Fires with AI Fault Prediction
Substation fires are rare, devastating — and almost always preceded by electrical warning signs that high-resolution waveform analysis can catch.
The next fault is already developing
The question is not whether it is coming — it is whether you find out 15 days early, or at the moment the breaker opens. Book a 30-minute demo with our founders, on real grid data.