Alarms after the fault
Threshold-based alarms only fire once damage is done. No prediction, little reaction time, long downtimes — and sometimes catastrophic failures.
Intelligent O&M for renewable energy
Predict failures before they happen.
PowerSense is an intelligent digital platform serving the whole life cycle of power electronics. Built on physics-informed machine learning, it monitors health, predicts faults and optimises maintenance across wind, grid, EV and storage infrastructure — using only the sensors you already have.
- Born at the University of Cambridge
- Validated on a commercial offshore wind farm
- Backed by MiraclePlus & supported by AWS
$50bn+
spent every year on renewable energy operation & maintenance
3×
more potential faults detected than conventional monitoring systems
1 week
of early warning before faults become failures
80%
O&M cost reduction shown in our cost–benefit validation
The problem
Maintenance today is passive — and expensive
Renewables already supply 30% of global electricity, yet the equipment behind them is still maintained reactively. Sensors raise alarms only after a fault has happened, and every fix means sending engineers to remote sites.
Manual troubleshooting
80–90% of O&M cost comes from corrective call-outs: engineers travelling to remote turbines and substations to diagnose problems by hand.
Lost generation
Offshore wind availability sits at just 95%. Untimely maintenance and undiagnosed faults cost the industry more than $15bn a year in lost production.
What we do
One engine across design, test and operation
PowerSense combines AI with physics to evaluate the multi-physics performance and reliability of power electronics — efficiently and accurately — at every stage of the product life cycle.
Design
Workflow-driven, multi-level design with an AI+physics engine and multi-objective optimisation — hundreds of times faster than conventional simulation.
How it works →Test
Reliability testing in a virtual environment under realistic conditions — finding design defects early and cutting testing time by 90%.
How it works →Operation & Maintenance
Real-time health monitoring, fault prediction, remaining-lifetime estimation and dynamic maintenance planning — with no additional sensors.
See the platform →The platform
From raw signals to a maintenance plan
A cloud software platform that deploys seamlessly into the control desks operators already use. It fuses existing sensor data, analyses performance, diagnoses faults, predicts remaining lifetime and plans the optimal maintenance schedule.
- No extra hardware — works with the sensors already installed by manufacturers
- Physics-informed AI — accurate insight even from low-quality raw data
- Full life cycle — from power modules and capacitors to whole converters
Field-proven
Validated on a commercial offshore wind farm
In collaboration with our partner, Offshore Renewable Energy Catapult, our prototype — the world’s first fault-detection solution for electrical converters — was validated on a commercial offshore wind farm.
It detected three times more potential faults than the existing monitoring system and prevented one catastrophic failure.
3×
more potential faults detected
80%
lower O&M cost in cost–benefit analysis
+2%
total power production
Backed by
MiraclePlus
Shenzhen Bao’an Doctorate Association Capital
Amazon Web Services
Partners & recognition
ORE Catapult
Cambridge Enterprise
Chris Abell Start-up Competition
2nd Prize
Awarded at the University of Cambridge competition sponsored by Cambridge Enterprise and Cambridge Innovation Capital.
“21 to Watch” 2023
Named
Shortlisted among the next generation of Cambridge game-changers poised for future success.
Global Start-up Competition, Shenzhen Bao’an
Runner-up
Runner-up among startups from around the world at the global competition held in Bao’an, Shenzhen.