Observe
Every movement produces a power signature that captures the physical behaviour of the device.
Case study · Predictive maintenance
From switch telemetry to a service decision.
Device-specific power signatures turn persistent deviations into a concrete maintenance recommendation.
Operating proof
7Unseen switchesUsed for out-of-sample verification
1:1Device baselineEvery switch is compared with its own norm
LiveService decisionDeviation becomes a specific intervention
02 / Problem
Periodic inspection could not see what changed between visits.
A switch can deteriorate after one scheduled inspection and before the next. The resulting failure restricts station capacity and creates operational cost, but raw telemetry alone does not tell a maintenance team what to do.
03 / System
An individual operating norm for every physical device.
The system analyses the power signature of each movement against that switch's own baseline. Persistent deviations are classified by the kind and urgency of work they indicate.
Compare
The new signature is evaluated against the individual norm, not a fleet-wide average.
Act
A sustained deviation becomes a service category, urgency and recommended intervention.
04 / Outcome
A model verified on devices it had never seen.
The model was tested on seven switches outside its training set. The client team confirmed its indications as symptoms of real faults, and the completed system was transferred to Kombud Group.