簡易檢索 / 詳目顯示

回結果列表
研究生： 科森勒
Mbhamali, Nkosenhle Andrew
論文名稱： 基於LLM解釋標準測量規則的混合BIM-AI框架實現自動工程量計算：以BIM資料在房間裝修的應用為例。
A Hybrid BIM–AI Framework for Automated Quantity Takeoff Using LLM-Based Interpretation of Standard Measurement Rules: A Case Study on Room Finishes Using BIM Data.
指導教授： 馮重偉
Feng, Chung-Wei
學位類別： 碩士
Master
系所名稱： 工學院 - 土木工程學系
Department of Civil Engineering
論文出版年： 2026
畢業學年度： 114
語文別： 英文
論文頁數： 148
中文關鍵詞： 建築資訊模型 、工程量清單 、檢索增強生成 、大型語言模型 、BIM-AI 、SSMBW 、裝修工程項目 、BIMVision
外文關鍵詞： Building Information Modelling, Quantity Takeoff, Retrieval Augmented Generation, Large Language Model, BIM-AI, SSMBW, Finishing Work Items, BIMVision
相關次數： 點閱：54  下載：5 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 工程量清單編制是建築成本管理的重要環節，因為它確定了估算、採購和專案成本控制所需的可計量數量。傳統的工程量清單編製方法通常耗時費力，且容易出現人為錯誤。另一方面，基於BIM的工程量清單編制高度依賴建築構件建模的完整性和準確性；因此，它無法量化未建模的工作項目數量（例如鷹架、黏合劑等）。這導致BIM數量計算與專業測量實踐之間存在差距。

    本研究發展了一個混合BIM-AI輔助的自動化工程量清單編制框架，重點在於採用南部非洲標準建築工程測量系統（SSMBW）的精裝修工作項目。該框架整合了SSMBW規則、基於本體的工作項結構化、檢索增強生成（RAG）、大型語言模型（LLM）解釋、BIM參數提取、BIM參數映射以及基於標準測量規則的工作項目數量生成。 RAG流程利用語意嵌入和FAISS向量索引檢索SSMBW結構化規則區塊，從而提高了規則選擇的準確性。

    本研究使用Revit 2024案例研究模型測試了該框架。 BIM資料透過pyRevit API腳本擷取並匯出為JSON格式。隨後，該框架將LLM所需的參數對應到可用的BIM字段，並將工作項目分類為已建模、推斷或未建模。案例研究結果表明，該框架成功產生了所選裝修工作項目的工程量清單。本研究使用兩個基準對框架和案例研究進行了評估：使用BIMVision自動工程量清單（QTO）評估基於BIM的工程量清單；並使用工程量測量師編制的工程量清單（BOQ）評估其與專業工程量測量邏輯的一致性。最終輸出包括以下工作項目：磁磚鋪設、天花板、玻璃安裝、油漆和鷹架。該框架還能夠排除不屬於內部裝修工作項目範圍的元素和項目，例如陽台和外部工作項目。研究結果表明，所提出的框架能夠將BIM提取的數據與人工智慧輔助的標準測量規則解釋相結合。本研究的主要貢獻在於開發了一種BIM-AI工作流程，該流程將SSMBW條款與BIM參數和最終工程量輸出關聯起來。

    Quantity take-off is an important stage in construction cost management because it determines measurable quantities required for estimating, procurement and project cost control. Traditional QTO methods are often time consuming, labor intensive and prone to human error. BIM-based QTO on the other hand is highly dependent on how completely and accurately building elements are modeled; therefore, it fails to quantify unmodeled work item quantities (e.g. scaffolding, adhesives etc.). This creates a gap between BIM quantity calculation and professional measurement practice.

    This research develops a hybrid BIM-AI assisted framework for automated QTO, focusing on finishing work items using Standard System of Measuring Building Works (SSMBW) used in Southern Africa. The framework integrates SSMBW rules, ontology-based work item structuring, Retrieval-Augmented Generation (RAG), Large Language Model (LLM) interpretation, BIM parameter extraction, BIM parameter mapping and standard measurement rule-based work item quantity generation. The RAG process retrieved SSMBW structured rule chunks using semantic embedding and a FAISS vector index, enhancing accurate rule selection.

    A Revit 2024 case study model was used to test the framework. BIM data was extracted using pyRevit API script and exported into JSON format. The framework then mapped the LLM required parameters to the available BIM fields and classified the work items as modeled, inferred or unmodeled. The case study results showed successful quantity generation for the selected finishing work items. The framework and the case study were evaluated using two benchmarks: BIMVision automated QTO was used to assess BIM-based QTO; and a quantity surveyor prepared BOQ was used to assess the alignment with professional quantity-surveying reasoning.The final outputs included the following work item trades; tiling, ceilings, glazing, paintwork and scaffolding. The framework also showed the ability to exclude elements and items that didn’t fall under the scope of internal finishing work items like the veranda and external work items. The results demonstrate that the proposed framework can combine BIM extracted data with AI-assisted standard measurement rule interpretation. The main contribution of the research is the development of a BIM-AI workflow that links SSMBW clauses to BIM parameters and final quantity outputs.

    ABSTRACT i–ii English Version i Chinese Version ii ACKNOWLEDGEMENT iii TABLE OF CONTENTS iv–vi LIST OF FIGURES vii–ix LIST OF TABLES x LIST OF SYMBOLS xi ABBREVIATIONS AND MEANINGS xii CHAPTER 1: INTRODUCTION 1–6 1.1 Introduction 1–2 1.2 Background and Motivation of Research 2–3 1.3 Objectives of Research 3–3 1.4 Research scope and limitations 3–4 1.5 Research Procedure 4–5 1.6 Thesis organization 6–6 CHAPTER 2: LITERATURE REVIEW 7–18 2.1 Problem Statement 7–11 2.1.1 Limitations of Traditional Quantity Takeoff 8–9 2.1.2 Limitations of BIM-based Quantity Takeoff 9–10 2.1.3 Difficulty of Applying Measurement Standards on BIM-QTO 10–10 2.1.4 Lack of AI Interpretation of Standard Measurement Rules 10–11 2.2 Literature Review 11–18 2.2.1 Traditional and BIM-Based Quantity Takeoff 11–13 2.2.2 Limitations of BIM-Based Quantity Takeoff for finishing works 13–13 2.2.3 Building Measurement Standards and BIM Integration 13–15 2.2.4 Artificial Intelligence and Reasoning in Quantity Takeoff 15–17 2.2.5 Hybrid BIM-AI Frameworks in Construction Automation 17–18 2.3 Summary 18–18 CHAPTER 3: RESEARCH METHODOLOGY 19–57 3.1 Tools and Software 19–26 3.1.1 Building Information Modelling (BIM) Tool 20–22 3.1.2 BIM Data Extraction Tool 22–23 3.1.3 Data Processing and Storage Tool 23–25 3.1.4 Programming Language and Development Environment 25–25 3.1.5 AI, RAG and Semantic Search Tools 26–26 3.2 Standard System of Measurement (SSMBW) Preparation 26–31 3.2.1 Selection of Applicable SSMBW Trades 27–27 3.2.2 Clause Structuring and Standardization 28–29 3.2.3 Clause Identifier Development and Classification 30–30 3.2.4 Measurement Rule and Deduction Requirement Extraction 30–31 3.3 Ontology and Semantic Knowledge Development 31–40 3.3.1 SSMBW Ontology Development 31–36 3.3.2 Work Item Classification Structure 36–38 3.3.3 Measurement Rule Relationships 38–38 3.3.4 BIM Parameter Requirement Relationships 38–40 3.4 LLM and RAG Configuration 40–44 3.4.1 Knowledge Base Preparation 40–41 3.4.2 Clause Embedding and Vector Indexing 41–41 3.4.3 RAG Workflow 41–42 3.4.4 Prompt Engineering Strategy 42–43 3.4.5 Structured Output Configuration 44–44 3.4.6 BIM Parameter Interpretation 44–44 3.5 BIM Data Extraction and Processing 45–49 3.5.1 Revit BIM Model Preparation 45–46 3.5.2 Room, Geometry, Parameter, Finish and Material Extraction 46–47 3.5.3 BIM Data Structuring and Export 47–48 3.5.4 BIM Data Cleaning and Normalization 48–49 3.6 AI-Assisted Standard Measurement Rule Interpretation and BIM based Quantity Takeoff 50–53 3.6.1 BIM Parameter Mapping 50–51 3.6.2 Quantity Logic Interpretation 51–52 3.6.3 Computable and Incomputable Work Items 52–52 3.6.4 Automated Quantity Report Generation 52–53 3.7 Framework Validation 53–57 3.7.1 Validation Methods 54–55 3.7.2 Accuracy Assessment 55–55 3.7.3 Completeness Assessment 56–56 3.7.4 Framework Limitations 56–57 CHAPTER 4: CASE STUDY AND RESULTS DISCUSSION 58–89 4.1 Case Study BIM Model 60–64 4.1.1 BIM Model Description 60–61 4.1.2 BIM Semantic Preparation 62–64 4.2 SSMBW Interpretation Results 65–71 4.2.1 Clause Interpretation Results 65–67 4.2.2 BIM Parameter Extraction Results 67–69 4.2.3 Quantity Logic Interpretation Results 70–71 4.3 BIM-AI based Generated Quantity Results’ Evaluation 71–83 4.3.1 Ceiling Quantity Results 74–75 4.3.2 Scaffolding Quantity Results 75–77 4.3.3 Paintwork Quantity Results 77–79 4.3.4 Tiling Quantity Results 79–81 4.3.5 Glazing Quantity Results 82–83 4.4 Validation and Discussion 84–89 4.4.1 Quantity Accuracy Assessment 84–87 4.4.2 Framework Performance 87–88 4.4.3 Limitations and Challenges 88–89 CHAPTER 5: CONCLUSIONS AND FUTURE DEVELOPMENT 90–93 5.1 Conclusions 90–92 5.2 Future Development 92–93 REFERENCES 94–96 APPENDICES 97–134

    Alathamneh, S., Collins, W. & Azhar, S., 2024. BIM-based quantity takeoff: Current state and future opportunities. Automation in Construction, Volume 165.
    Alshabab, M., Vysotskiy, A., Petritchenko, M. & Khalil, T., 2020. BIM-Based Quantity Takeoff in Autodesk Revit and Navisworks Manage. Lecture Notes in Civil Engineering, Issue 70, pp. 413-421.
    ASAQS, 2021. Standard System of Measuring Building Work. 7th ed. Pretoria: Association of South African Quantity Surveyors.
    Autodesk Revit, 2025. Revit API Developers Guide. [Online] Available at: https://help.autodesk.com/view/RVT/2027/ENU/?guid=Revit_API_Revit_API_Developers_Guide_Basic_Interaction_with_Revit_Elements_Parameters_html [Accessed 3 June 2026].
    Construction Management Association of America (CMAA), 2025. Adoption of Artificial Intelligence in Quantity Takeoffs in Engineering & Construction Organizations. NAC Executive Insights.
    Fürstenberg, D. et al., 2024. Automated quantity take-off in a Norwegian road project. Scientific Reports, Volume 14.
    Gao, Y. et al., 2023. Retrieval-Augmented Generation for Large Language Models: A Survey. Computation and Language (cs.CL), Issue arXiv preprint, arXiv:2312.10997.
    Garner, J. & Soornack, Y., 2025. Quantity Surveying x AI: current state. [Online] Available at: https://www.quantik.co.uk/insights/articles/quantity-surveying-x-ai-current-state/ [Accessed February 2026].
    Glodon International, 2025. How to automate quantity takeoffs: A deep dive into AI & BIM integration. [Online] Available at: https://www.youtube.com/watch?v=JQ2rz1nYcvg [Accessed February 2026].
    Han, J., Lu, X. & Lin, J., 2025. Unified Network-Based Representation of BIM Models for Embedding Semantic, Spatial, and Topological Data. Computational Engineering, Finance, and Science, Issue 42.
    Liu, H., Cheng, J. C., Gan, V. J. & Zhou, S., 2022. A knowledge model-based BIM framework for automatic code-compliant quantity take-off. Automation in Construction, Volume 133.
    Liu, H., Cheng, J. & Gan, V., 2021. Developing a knowledge-based system for semantic enrichment and automatic BIM-based quantity take-off. Proceedings of the 38th ISARC, pp. 169-175.
    Liu, H., Cheng, J., Gan, V. & Zhou, S., 2022. A novel Data-Driven framework based on BIM and knowledge graph for automatic model auditing and Quantity Take-off. Advanced Engineering Informatics, Volume 54.
    Li, Y., Zhang, J. & Wang, X., 2023. Leveraging large language models for automated construction document interpretation. Journal of Construction Engineering and Management, 149(8).
    Meta AI, 2026. Faiss. [Online] Available at: https://ai.meta.com/tools/faiss/ [Accessed May 2026].
    National Academy of Construction (NAC), 2025. Quantity Takeoffs in AI Enabled Digital Twins. Executive Insights.
    OpenAI, 2024. Structured Model outputs. [Online] Available at: https://developers.openai.com/api/docs/guides/structured-outputs [Accessed May 2026].
    Palcode, 2025. Top Construction Quantity Takeoff Software: 2025 ROI Comparison. [Online] Available at: https://palcode.ai/top-construction-quantity-takeoff-software-2025-roi-comparison/ [Accessed February 2026].
    Pandas, 2026. Pandas Documentation. [Online] Available at: https://pandas.pydata.org/docs/ [Accessed May 2026].
    Royal Institution of Chartered Surveyors (RICS), 2025b. Responsible use of artificial intelligence in surveying practice. 1st ed. London: Royal Institution of Chartered Surveyors (RICS).
    Royal Institution of Chartered Surveyors (RICS), 2025. RICS artificial intelligence in construction report 2025. [Online] Available at: https://www.rics.org/news-insights/artificial-intelligence-in-construction-report [Accessed February 2026].
    Saenpaeng, W., Tantrairatn, S. & Chaphirom, A., 2025. Development of an Automated Workflow for Reinforced Concrete Structural Quantity Takeoff and Cost Estimation Using Visual Programming in a BIM Environment. Engineering and Technology Horizons, 42(4).
    SBERT.net, 2026. Semantic Textual Similarity. [Online] Available at: https://sbert.net/docs/sentence_transformer/usage/semantic_textual_similarity.html [Accessed May 2026].
    Standards South Africa, 2004. SOUTH AFRICAN NATIONAL STANDARD (SANS 10085-1:2004). In: The design, erection, use and inspection. Pretoria: Standards South Africa.
    Taghaddos, H., Mashayekhi, A. & Sherafat, B., 2018. Automation of Construction Quantity Take-off in a BIM Model. Computational Engineering, Finance, and Science.
    The Association of South African Quantity Surveyors, 2015. Standard System of Measuring Building Work. 7th ed. s.l.:The Association of South African Quantity Surveyors.
    Valdebenito, R. & Eric Forcael, E., 2025. Integrating Artificial Intelligence and BIM in Construction: Systematic Review and Quantitative Comparative Analysis. Applied Sciences, 15(23).
    Valinejadshoubi, M. et al., 2024. Automated system for high-accuracy quantity takeoff using BIM. Automation in Construction, Volume 157.
    Valinejadshoubi, M. et al., 2025. A Cloud-Driven Framework for Automated BIM Quantity Takeoff and Quality Control, Buildings. 15(21).

    下載圖示
    校外：立即公開
    QR CODE