Digital twins: fidelity, synchronisation and the validation gap¶
Scope and how to read this survey¶
brief: Framing only. State the reader, the three themes, and the exclusions from scope.md in two short paragraphs. No citations here -- the exclusions are ours to declare, not the literature's.
What the field means by "digital twin"¶
brief: Establish that the term is contested before anything rests on it. Three or four definitional families, each with its strongest proponent. Do not adjudicate; the survey's own working definition is in the glossary and belongs in the framing section above.
queries:
- digital twin definition
- digital twin taxonomy classification
- digital shadow versus digital twin
Fidelity is chosen, not maximised¶
claim: Higher model fidelity is not uniformly better. It is chosen against the decision the twin supports, and a more faithful model that misses its deadline is worse than a coarse one that does not.
claim: The field has no shared threshold for "faithful enough", which is why fidelity claims across papers are not comparable.
queries:
- digital twin model fidelity
- surrogate model accuracy tradeoff
- real time simulation deadline fidelity
Keeping the twin in step with the asset¶
brief: The data half. Sampling rate, staleness, and what each paper does when the link drops -- that last one is where the corpus is most practical and least cited.
queries:
- digital twin data synchronisation latency
- sensor sampling rate state estimation
- intermittent connectivity state reconstruction
Comparison of the surveyed approaches¶
brief: This is the table, not prose. One row per approach, columns for fidelity treatment, synchronisation strategy, validation performed, and domain. A row that repeats a sentence from above should lose the sentence, not the row.
Validation, and why it is mostly absent¶
claim: Most of this corpus validates against another model rather than against the physical asset, and the papers that do validate against an asset do so on a single installation.
brief: Be blunt. This is the gap the reader's thesis can occupy, and softening it wastes the survey's most useful paragraph.
queries:
- digital twin validation verification
- model validation physical experiment
- simulation credibility assessment
Gaps and what would close them¶
brief: Three or four, each phrased as a question a study could answer, each traceable to a thin spot named above. No new citations -- if a gap needs a source the reader has not met yet, it belongs in an earlier section.