The solid shaft (left) is the physical asset. The translucent wireframe (right) is its digital twin — a virtual model driven by delayed, noisy sensor telemetry rather than direct physics access.
vibration(t) = k · imbalance · ω²
twin_state(t) = physical_state(t - latency) + noise
error = |twin_state - physical_state| / physical_state
- Shaft speed (RPM) — angular velocity ω of the real machine; vibration scales with ω².
- Mass imbalance — an eccentric mass on the disk that produces synchronous vibration, modelling wear or a manufacturing defect.
- Sensor noise — random error added to telemetry before it reaches the twin model.
- Twin sync latency — communication/computation delay between the physical asset and its virtual replica.
- Inject Fault — suddenly increases imbalance, testing how fast the twin detects the anomaly through its lag.
Real application: manufacturers run digital twins of turbines and gearboxes to predict bearing wear and schedule maintenance before physical failure, using exactly this kind of delayed sensor-driven virtual model.