What is anomaly detection?

Flagging behaviour that doesn't match an asset's normal pattern.

Anomaly detection learns what 'normal' looks like for an asset — its vibration envelope, temperature range, current draw — and flags readings that drift outside it, without needing a pre-defined fault signature.

It catches the faults nobody modelled: instead of alarming only on known failure modes, it surfaces 'this machine changed'. Detected anomalies route to a human for diagnosis and work order.

• How it's used