數位孿生的定義與分類學，以機器學習與科學應用案例研究A definition and taxonomy of digital twins: case studies with machine learning and scientific applications
Weingram et al.｜Frontiers in High Performance Computing 3: 1536501｜DOI: 10.3389/fhpcp.2025.1536501
狀態：AI_DRAFT_FROM_REVIEW｜分級：B｜閱讀深度：ABSTRACT_CROSSREF_CHECKED｜Jacky 審核：False
通用數位孿生底層致能
digital twindefinitiontaxonomymachine learningHPCtaxonomycase studyphysics-informedglobal
專討核心文獻定位
[23]
DT · 跨產業數位孿生
新增
建立一致的數位孿生定義與機器學習強化的分類學，並以科學應用案例研究佐證
使用警示
本頁是文獻知識庫卡片，不等於可直接引用的最終查核稿。只有狀態升級為 CITABLE 後，才可直接進入論文引用候選。
為什麼納入這篇
This paper is a bottom-layer foundational source that nails down what a digital twin actually is, and proposes a taxonomy for machine-learning-enhanced digital twins across scientific applications, which anchors the cross-industry definition layer of the knowledge base.
結構式摘要｜中英文對照
| 研究問題 | 數位孿生應如何被一致地定義，以及機器學習強化的數位孿生應如何被分類，才能在科學計算與高效能運算情境下被清楚理解與應用？ How should digital twins be defined consistently, and how should machine-learning-enhanced digital twins be classified, so that they can be clearly understood and applied in scientific computing and high-performance computing contexts? |
|---|---|
| 資料來源 | 摘要確認：本文針對下一世代科學儀器與模擬所產生的超大資料集，論述需要 HPC 技術做即時且準確的分析，並以此釐清數位孿生、科學計算與機器學習三者的關係。具體資料集規模與案例細節待查。 The abstract confirms that the paper addresses the very large datasets produced by next-generation scientific instruments and simulations, argues that HPC techniques are needed for real-time and accurate analysis, and uses this to clarify the relationship among digital twins, scientific computing, and machine learning. Specific dataset sizes and case details are to be checked. |
| 方法 | 摘要確認：本文建立一致的數位孿生定義，並提出機器學習強化數位孿生的分類學（taxonomy），再以案例研究說明。具體分類維度數量與案例數待查。 The abstract confirms that the paper establishes a consistent definition of digital twins and proposes a taxonomy of machine-learning-enhanced digital twins, illustrated through case studies. The exact number of taxonomy dimensions and case studies is to be checked. |
| 主要結果 | 摘要確認：本文給出一致的數位孿生定義與 ML 強化的分類學，並指出未來方向包含 hybrid assimilation 框架與 physics-informed 技術以提升準確度。具體數字、Table 與 Figure 編號待查。 The abstract confirms that the paper provides a consistent definition and an ML-enhanced taxonomy, and points to future directions including hybrid assimilation frameworks and physics-informed techniques to improve accuracy. Specific numbers, table, and figure identifiers are to be checked. |
| 限制 | 待查。本卡僅依摘要與 Crossref 書目查核撰寫，尚未讀全文，限制與案例邊界需讀全文後補。 To be checked. This card is written from the abstract and Crossref bibliographic verification only; the full text has not been read, so limitations and case-study boundaries must be added after full-text reading. |
Key Findings
| 發現 | 證據 | 確定性 |
|---|---|---|
| The paper establishes a consistent definition of digital twins and a taxonomy of machine-learning-enhanced digital twins. | Abstract states the work builds a consistent definition and a taxonomy for ML-enhanced digital twins, with case studies. | checked_against_abstract_and_crossref |
| HPC techniques are positioned as necessary for real-time, accurate analysis of very large scientific datasets, linking digital twins, scientific computing, and machine learning. | Abstract frames next-generation instruments and simulations as producing very large datasets that require HPC for real-time accurate analysis. | checked_against_abstract_and_crossref |
| Future directions include hybrid assimilation frameworks and physics-informed techniques to improve accuracy. | Abstract names hybrid assimilation and physics-informed techniques as future directions. | checked_against_abstract_and_crossref |
Key Figures and Tables
公開網站原則：未確認授權前，不直接複製原文圖表；優先使用自製圖表導讀或重繪圖。
| 項目 | 內容 | 關鍵數字 | Jacky 判讀 | 重用策略 |
|---|---|---|---|---|
| 待查 | 待查；尚未讀全文，無法確認 Figure / Table 內容。 | 待查 | 讀全文後，可考慮以本文的分類學維度作為跨產業數位孿生底層定義的參照表。 | CC-BY 授權雖允許重用，仍待逐圖確認後再決定是否重繪或引用。 |
Extracted Evidence Table
| 可支撐主張 | 指標或結果 | 原文位置 | 可引用 | 備註 |
|---|---|---|---|---|
| Digital twin can and should be given a consistent definition, with an explicit taxonomy for ML-enhanced variants. | Abstract: a consistent definition plus a taxonomy of ML-enhanced digital twins, supported by case studies; specific counts 待查. | Abstract (full-text locations 待查). | True | Cite as a definition-and-taxonomy foundation source; do not attach specific numbers, tables, or figure identifiers until full-text checked. |
Critical Appraisal
Strengths
- Directly tackles the definition problem, which is foundational for the whole knowledge base.
- Proposes an explicit taxonomy rather than yet another ad hoc description.
- Bridges digital twins, scientific computing, machine learning, and HPC.
Weaknesses
- Card written from abstract and Crossref only; specifics 待查.
- Generality across scientific domains may not directly map onto forest use cases without adaptation.
| Validation quality | 待查 until full text checked |
|---|---|
| Transferability to Taiwan | medium as a definitional reference; not domain-specific to Taiwan |
| Risk of overclaiming | Do not present any specific dataset size, taxonomy dimension count, or case-study result as confirmed until the full text is read; abstract supports only the definition, taxonomy, and stated future directions. |
與 Jacky 博論 / Review 的用途
| 博士論文 | Supports the dissertation's need for a defensible, consistent definition of digital twin and a taxonomy lens for ML-enhanced twins, sitting at the bottom definitional layer beneath forest-specific work. |
|---|---|
| TJFS Review | Gives the TJFS review a cross-industry definitional anchor so that forest digital twin is framed against an agreed definition rather than treated as a 3D visualization product. |
| 可引用句候選 | 2025 年，Weingram 等人發表的文獻中指出，數位孿生需要一致的定義與機器學習強化的分類學，才能在科學計算與高效能運算情境下被清楚理解。 |
| 不可用來主張 | Do not use this paper for forest-specific empirical numbers, nor cite any internal figure, table, or dataset size before the full text is verified. |
授權與圖表重用
| Article license | CC-BY |
|---|---|
| Figure reuse policy | DO_NOT_REUSE_ORIGINAL_FIGURES_PUBLICLY_UNTIL_LICENSE_CHECKED |
| Notes | Article is CC-BY 4.0; full-text translation permitted, but figure reuse held until each figure's license is individually verified. |
待查核清單
- 讀全文，補上分類學的維度數、案例研究數與具體 Table / Figure 編號。
- 確認摘要以外的數據與限制，將 read_depth 升級至 FULL_TEXT_CHECKED。
- CC-BY 授權下，逐圖確認後再決定是否重繪或引用原圖。
- 等 Jacky 人工審核後再考慮列入可引用候選。