Unbalance
Uneven mass distribution on the rotor. Dominant at running speed, and one of the most common — and most correctable — causes of excess vibration.
Predictive Maintenance · UK
AI Vibration Monitoring & Predictive Maintenance
Wireless vibration sensors and edge AI that recognise unbalance, misalignment and bearing wear while they are still minor — and turn a changing signature into a planned maintenance job instead of an unplanned stoppage.
A 30-minute technical conversation, not a sales pitch.
Pump 3 · velocity spectrum
2× running speed rising
[ 01 ] WHAT IT DETECTS
Six faults, six different signatures
Every developing mechanical fault changes how a machine vibrates, and each changes it in a recognisable way. That is what lets a model do more than say "something is wrong" — it can tell you what, and how urgently.
Shaft misalignment
Often confused with unbalance, but reads differently in the spectrum. How to tell the two apart.
Bearing wear
Inner-race, outer-race and rolling-element defects progress through recognisable stages. The five stages of bearing failure.
Mechanical looseness
Worn mountings, soft foot and loose fasteners produce a spread of harmonics that grows as the looseness does.
Gear wear
Tooth wear and damage show as sidebands around the gear-mesh frequency long before they are audible on site.
Pump cavitation
Collapsing vapour bubbles raise random high-frequency energy — an operating-condition problem as much as a mechanical one.
[ 02 ] HOW IT WORKS
From raw vibration to a work order
The same four stages whether you are monitoring five pumps or five hundred — only the scale of the infrastructure changes.
Sense
Wireless tri-axial vibration and temperature sensors on critical rotating assets — or your existing sensors and historian data where they already cover the machine.
Process at the edge
A gateway beside the machine computes spectra and features locally, so only the useful information travels — and monitoring continues through network outages.
Detect & diagnose
Models learn each asset's normal behaviour, flag deviations, and classify the likely fault so an alert arrives with a probable cause, not just an amplitude.
Act
Alerts land in your CMMS as work orders, on dashboards, or on a live Digital Twin of the asset — with guardrails so nothing safety-critical happens without a person approving it.
[ 03 ] WHAT WE MONITOR
Rotating equipment where downtime costs most
We start with the assets whose failure stops production, and extend coverage only once detection has proved itself on your plant.
MotorsPumpsFans & blowersCompressorsGearboxesConveyorsCrushersMixers & agitators
What you get
- An asset criticality review — which machines to monitor first, and why.
- A sensor and connectivity design suited to your site: LoRaWAN, Wi-Fi or wired, with power and gateway placement planned.
- Anomaly detection and fault classification trained on your own machines' baselines.
- Integration with the systems you already run — SCADA, PLCs, historians and your CMMS over OPC-UA, Modbus or MQTT.
- Dashboards and alerting your maintenance team will actually use, plus handover training.
[ 04 ] HOW WE ENGAGE
Prove it on a few machines before you scale
Technical audit
We review your critical assets, existing data and connectivity, and tell you plainly whether predictive maintenance is worth it for your site.
Prove detection
A scoped pilot on a handful of machines, with agreed success criteria, so you see real detections on your own equipment before committing to a rollout.
Roll out & hand over
Extend coverage across the site, connect it to your maintenance workflow, and train your team to own it — with ongoing model support if you want it.
[ 05 ] FURTHER READING
Technical guides from our engineers
Guide
Predictive Maintenance: The Complete Guide
Read → DiagnosticsUnbalance vs Misalignment: Why They Look Similar but Aren't
Read → DiagnosticsBearing Fault Progression: 5 Stages Every Reliability Team Should Know
Read → AIHow AI Interprets Vibration Data
Read → AIEarly Fault Detection with AI Vibration Analysis
Read → WorkshopApplied AI for Engineers — anomaly detection on real sensor data
See dates →
[ 06 ] FAQ
Common questions
What faults can vibration monitoring detect?
Vibration analysis reliably detects unbalance, shaft misalignment, mechanical looseness, rolling-element bearing defects, gear wear and, on pumps, cavitation. Each fault leaves a characteristic signature in the frequency spectrum, which is what lets a model tell them apart.
Do we need to replace our existing sensors or SCADA system?
No. Where suitable sensors or historian data already exist we use them, and we integrate alongside SCADA, PLCs and your CMMS over standard protocols such as OPC-UA, Modbus and MQTT. New wireless sensors are only added where there is a monitoring gap.
Why run the AI at the edge rather than in the cloud?
High-frequency vibration data is expensive to stream continuously. Processing it on an edge gateway next to the machine sends only features and alerts upstream, cuts bandwidth, keeps working through connectivity outages and reacts in seconds rather than minutes.
How do you start a predictive maintenance project?
With a free technical audit of your critical assets, followed by a scoped pilot on a small number of machines. The pilot proves detection on your own equipment before any wider rollout is committed to.
Which of your machines would you least like to fail next month?
Start there. Tell us about the asset and we'll tell you honestly whether vibration monitoring would have caught its last failure — and what it would take to catch the next one.