Architecture

How Watsynq Works

Three heterogeneous sensor streams. One physics-informed fusion model that outputs a Remaining Useful Life score with uncertainty bounds.

Close-up of a vibration and acoustic emission sensor clamp mounted on a bearing housing of an industrial motor

Clip-on sensors. No process stop.

Acoustic emission clamps on the housing exterior. Oil particle counter inline-taps the existing lube circuit. Surface thermal sensor adheres to housing with thermal paste.

  • AE sensor clamps on bearing housing exterior — no bore-in fitting required
  • Oil particle counter taps into existing lube line — no process shutdown
  • Surface thermal sensor adheres with thermal paste — 5-minute install
  • Typical install per monitoring point: under 4 hours

Bearing Assembly — Sensor Placement

AE THERM OIL 150 kHz Rate-of-change ISO 4406

From edge sensor to RUL score.

Encrypted transmission, timestamp normalisation, signal alignment, and stage-adaptive physics weighting — all in the cloud pipeline. No data science team on-site. No custom model maintenance by the plant.

Edge Device
3-stream local aggregation
TLS 1.3 Transit
Encrypted uplink
Cloud Ingestion
Timestamp normalisation
Signal Alignment
Multi-stream sync
Failure Classifier
180+ defect signatures
Physics Model
Bayesian posterior
RUL Output
Score + CI to CMMS/API

Not just ML. Physics-informed fusion.

The model knows which signal matters at which failure stage — and adjusts weighting accordingly.

Failure mode library: 180+ bearing, gear, and impeller defect signatures indexed by equipment class. The model matches your equipment type before it starts scoring.

Stage-adaptive weighting: Acoustic emission is weighted highest in early-stage defect where stress waves precede structural damage. Thermal weighting increases in advanced stages where bearing friction raises housing temperature asymmetrically.

Uncertainty quantification: Bayesian posterior on RUL — not a point estimate. You see the confidence interval and the underlying signal contributions at every update.

Signal weight by defect stage

Signal
Early stage
Late stage
Acoustic
HIGH
MED
Thermal
LOW
HIGH
Oil
MED
MED

What you see and what it means — line by line.

Every alert has enough context for a maintenance planner to act without opening a second tool.

WATCH — Early stage defect detected
Equipment Kiln Fan #3 — Inboard Bearing
Location Line 2 / Bay 4
Failure mode Inner race defect — stage 1
RUL estimate 38 days
95% CI 29 – 49 days
Top signal Acoustic emission (62%)
Recommended action Schedule bearing replacement in next outage window before day 29
Sensor quality All sensors nominal
Last updated 07:14 today

Ready to see a live RUL output for your equipment type?

We will walk through the physics, show you a real alert output, and map installation to your specific equipment class. No slides.