Digital Twin

When to use this

Consult the digital twin when you need a complete view of an asset's state in one place -- from live sensor data to health predictions. Think of it as the asset's medical record: all its relevant information in a single view.

Real-world example: The shift operator receives a high-vibration alert on Pump C-12. Instead of checking multiple screens, they open the digital twin and see at a glance:

• L1 (Acquisition): The vibration sensor reports 12.5 mm/s (last reading 30 seconds ago)
• L2 (Trends): Vibration has increased 40% over the last 7 days
• L3 (State): Condition is "unsatisfactory" with a score of 0.35
• L4 (Health): AHI grade D, alarm health sub-index in red
• L5 (Prognostics): Estimated remaining useful life: 12 days at the current degradation rate
• L6 (Advisories): AI recommends bearing inspection and shaft alignment

With this information, the operator can make an informed decision: schedule preventive maintenance before the pump fails.

What is a digital twin

A digital twin is a complete virtual representation of a physical asset. In Rela AI, the digital twin aggregates information from the six layers of the ISO 13374 standard into a unified view, allowing you to understand the current, historical, and future state of equipment from a single endpoint.

The 6 ISO 13374 layers

The digital twin organizes information into six layers, each providing a progressive level of analysis:

Digital twin structure

Asset information

Basic asset data: name, code, type, manufacturer, model, serial number, criticality, status, and location.

L1 — Data Acquisition

• Latest values received from each event source.
• Timestamp of the last reading.
• Connection status of each source.

L2 — Data Manipulation

• Rolling statistics per time window (min, max, average, standard deviation).
• Trends with linear regression (slope, r_squared).
• Moving averages (SMA and EMA).
• Detected anomalies.

L3 — State Detection

• Current baseline and operating mode.
• Condition evaluation: good, acceptable, unsatisfactory, unacceptable.
• Deviations per metric.
• Condition score (0.0 - 1.0).

L4 — Health Assessment

• Current AHI and grade (A/B/C/D/F).
• Sub-index breakdown: condition, alarm_health, maintenance_compliance, trend_stability.
• Snapshot history.
• Deterministic recommendations.

L5 — Prognostics

• Estimated remaining useful life (RUL).
• Degradation rate (AHI points/day).
• Condition-based maintenance (CBM) triggers.
• Risk level.

L6 — Advisories

• AI-generated recommendations (Gemini).
• Lifecycle of each recommendation (open, acknowledged, resolved).
• Confidence based on data availability.

Maintenance plans

The digital twin includes maintenance plans linked to the asset:

• Scheduled preventive tasks.
• Corrective maintenance history.
• Next planned interventions.
• Plan compliance (percentage).

{
  "maintenance_plans": [
    {
      "name": "Monthly lubrication",
      "frequency": "monthly",
      "next_due": "2026-04-01",
      "compliance": 95.0
    }
  ]
}

Single endpoint

The digital twin is available from a single API endpoint:

GET /api/v1/assets/{asset_id}/digital-twin

Use cases

• Quick diagnosis — Get a complete view before making maintenance decisions.
• Executive reports — Generate health summaries of critical assets.
• CMMS integration — Feed maintenance management systems with structured data.
• Audits — Document the historical state of equipment.

See also

• Asset Management — Create and configure assets before viewing their digital twin
• Asset Health — Details on the health index (AHI) and sub-indices
• Prognostics — Remaining useful life and degradation rate
• Maintenance Plans — Schedule the interventions shown in the digital twin
• Event Sources — Connect the sensors that feed layers L1-L6