Skip to the content ForoHackers
Log inJoin

Digital twins: start here

Digital twins
#375 · Ophuscado · · Report

A digital twin is a living model of a real asset, such as a machine, a building or a plant, kept in step with it by data from its sensors. It lets you ask what is happening, what will happen and what would happen if you changed something, without touching the real thing.

Key concepts

  • Model plus data link: a simulation or data-driven model, fed by measurements so it follows its twin.
  • Fidelity: how closely the model matches reality; more detail costs more to build and to run.
  • State estimation: inferring quantities you cannot measure from those you can.
  • Uses: monitoring, fault diagnosis, forecasting wear and testing changes in a safe copy.
  • Lifecycle: the twin is calibrated, validated and updated as the real asset ages and changes.

Vocabulary

  • Physics-based model: built from equations of the system's behaviour.
  • Data-driven model: learned from recorded measurements.
  • Calibration: adjusting model parameters until it matches observations.
  • Co-simulation: running several models, each in its own tool, together.

What to study, in order

  1. Simulation and modelling of a simple system, such as a tank or a motor.
  2. Collect time-stamped data from a real or simulated source.
  3. Calibrate the model against the data and measure the error.
  4. Add state estimation so that the model follows live measurements.
  5. Use the twin to detect a fault and to test a change before applying it.

Related tags

#simulation #industrial-iot #control-systems #physics-simulation

Log in or join to reply.