在不動產開發先期規劃的階段決策建築量體時，建築設計者除了考量設計條件外也須與其他單位配合進行機能規劃及財務假設，以電腦軟體做為輔助來增加作業效率的方式已行之有年。然而在方案反覆修正及與他人溝通的過程中，兼顧修正需求時之餘仍需不斷以人力檢討法規限制及土管規定，傳統的作業方式由設計者人力進行配置決策除了費時費力外，亦無法即時且有系統性的判斷方案的優劣，因此本研究以衍生式設計建立一套建築量體配置決策系統，模擬在台北市內之基地進行量體規劃。透過整理出法規中量體相關的條文做為限制及設計建築量體生成的邏輯來產生量體方案，並藉由開放街圖(OSM)、 Ladybug 等軟體取得環境資訊，衍生為評斷方案優劣的依據，再將上述條件帶入多目標化式計算中迭代優化生成方案。演算後視不同目的透過篩選適存值，或以分群法分類以篩選出合適的方案。透過本架構進行量體配置與決策，可以泛化的應對都市地區多數基地，設計者僅需輸入基本資訊和做出少許規劃即可系統性的優化及產生大量方案。在設計產出的過程中反覆檢討的工作由電腦取代，有利於設計者將精力花在探討更有意義的議題上。

While doing building massing at the pre-planning stage of a real estate development, architecture designers not only need to consider the design conditions, but also need to cooperate with other professional teams to carry out functional planning and financial assumptions. Nowadays, computer software’s has been widely used to increase work efficiency; however, we still spend a lot of time and human resources on communicating with each profession to correct the drawings and regulation’s review. In tradition, it takes time and manpower to do the massing strategy all by the designers themselves, and it is hard to judge the pros and cons of the plan systematically. Therefore, this study constructs a “building massing strategy system” by generative design method and evolutionary algorithm, to stimulate a massing planning of a site in Taipei. By sorting out the regulations related to the site restriction and obtain the environmental information by software such as Open Street Map (OSM), Ladybug, etc., to derive as the basis of pros and cons, we could bring into evolutionary algorithm to optimize the massing proposal, then filter out which proposal suit the best to the site by different depending purpose and fitness value after the simulation. Through this massing tool, we could easily deal with most part of sites in the urban district, the designer only need to input the basic information and few criteria of the site, then can get various type of massing proposals. The repeating reviewing task by human has been replace by computer, which provide more time for the designers to handle other meaningful issues.

一、中文文獻
[1] 蔡宗瑋 (2013)，性能導向之參數式設計於 BIM 整合評估流程，碩士論文，國立成
功大學建築系。
[2] 林豐澤 (2005)，演化式計算上篇：演化式演算法的三種理論模式，智慧科技與應
用統計學報, 3(1), 2005。
[3] 吳獻堂 (2000)，人工智慧多目標最佳化程式模組系統之開發－以基因演算法為基
礎(II)，行政院國家科學委員會專題研究計畫成果報告。
[4] 蕭智夫 (2005)，都市設計建築量體管制原則之研究－以新市區建設地區為例，碩
士論文，國立成功大學建築系。
[5] 南方英 (2012)，電腦輔助建築量體設計之方法，碩士論文，國立台灣大學土木工
程系。
[6] 劉秉承 (2016)，『形隨評析』- 以空間型構理論與形態運算衍生為基礎之參數式設
計方法，臺灣建築學會「建築學報」第 96 期增刊（建築人文空間專刊），27~49
頁。
[7] 彭思齊 (2013)，模矩式集合住宅參數化之研究，碩士論文，淡江大學建築系。
[8] 湯天維 (2012)，以演算法探討參數化集合住宅設計，碩士論文，淡江大學建築
系。
[9] 衛萬明、 闕伯勳 (1999)，應用遺傳演算法於建築設計平面空間配置之研究，碩士
論文，朝陽科技大學建築及都市設計研究所。
[10] 張瑋晟 (2015)，衍生式設計工具在設計初期應用於法規檢測與量體優化，碩士論
文，國立台灣科技大學建築系。
二、英文文獻
[1] Jones, J. C. (1970), Design Methods: Seeds of Human Futures, London, UK: WileyInterscience Press.
51
[2] Schumacher P (2012), The Autopoiesis of Architecture. Vol. II: A New Agenda for
Architecture. London, John Wiley & Sons.
[3] Gibson,E. J. (1982).Working with the Performance Approach in Building. CIB Report
Publication 64. Rotterdam,The Netherlands. CIB (International Council for Research and
Innovation in Building and Construction).
[4] Frazer, J. (2002) Creative Design and the Generative Evolutionary Paradigm, The
Morgan Kaufmann Series in Artificial Intelligence, Pages 253-274 5.
[5] Oxman, R. (2006), Theory and design in the first digital age, Design Studies, 27 (3)
(2006), p. 229 .
[6] Goldberg D. E. (1989), Genetic Algorithms in Search, Optimization and Machine
Learning, Addison Wesley Publishing Co. Inc.
[7] Nauata, N., Chang, K.-H., Cheng, C.-Y., Mori, G., Furukawa, Y.. (2020). F House-GAN:
Relational Generative Adversarial Networks for Graph-Constrained House Layout
Generation. Lecture Notes in Computer Science (including subseries Lecture Notes in
Artificial Intelligence and Lecture Notes in Bioinformatics)12346 LNCS, p.162-177
[8] Hillier, B., & Hanson, J. (1984). The Social Logic of space. Cambridge, UK: Cambridge
University Press. 8
[9] IWeixin Huang , Hao Zheng.(2018).Architectural Drawings Recognition and Generation
through Machine Learning at the 38th Annual Conference of the Association for
Computer Aided Design in Architecture (ACADIA), Mexico City, Mexico, October
2018