請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88492
標題: | 基於MMC方法之混合式拓樸最佳化架構應用於撓性機構設計 MMC-based Hybrid Topology Optimization Framework for Compliant Mechanism Design |
作者: | 鄧名宇 Arthur Teng |
指導教授: | 李志中 Jyh-Jone Lee |
關鍵字: | 撓性機構,最佳化設計,MMC拓樸最佳化,啟發式演算法,非線性有限元素分析, Compliant Mechanism,Optimal Design,MMC Topology Optimization,Metaheuristic Algorithm,Nonlinear Finite Element Analysis, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 本研究提出一個全新的拓樸最佳化(Topology Optimization)架構,稱為基於MMC(Moving Morphable Component)方法之混合式拓樸最佳化,用於設計撓性機構(Compliant Mechanism)。此架構以MMC拓樸最佳化為基礎,融合了啟發式演算法(Metaheuristic Algorithm)和梯度式演算法(Gradient-based Algorithm)。由於撓性機構的運動通常涉及大形變,演算法使用非線性有限元素分析以獲取準確的結構響應。混合式拓樸最佳化分為三個階段:啟發式搜索(Metaheuristic Search)、部件擴增(Component Augmentation)和梯度式搜索(Gradient-based Search)。首先,啟發式搜索利用極少量的B樣條尋找機構骨幹。接著,部件擴增將機構骨幹轉換為梯度式搜索的初始猜值。最後,梯度式搜索找出機構的最佳拓樸。本研究透過三個設計問題展示混合式拓樸最佳化的效用。此外,本研究也針對混合式拓樸最佳化的效能做進一步的分析,包括運算加速、對特定參數的敏感度和初始猜值的影響。本研究提出的架構不僅能消除拓樸最佳化對人為設定初始猜值的依賴性,還能夠高效率地運行啟發式搜索。 This study proposes a novel topology optimization framework called MMC (Moving Morphable Component)-based Hybrid Topology Optimization for compliant mechanism design. Based on MMC Topology Optimization, this framework incorporates both metaheuristic and gradient-based algorithms into the optimization process. Due to the fact that compliant mechanisms usually undergo large deformation, the algorithms employ nonlinear finite element analysis to obtain accurate structure responses. The Hybrid Topology Optimization consists of three main stages: Metaheuristic Search, Component Augmentation, and Gradient-Based Search. In the first stage, a small number of splines are used to search for the backbone structure of the mechanism. Then, in the second stage, the backbone structure is converted into an initial guess for the next stage. Finally, the algorithm identifies the optimized topology in the third stage. This study demonstrates the effectiveness of the proposed framework through three design problems. Furthermore, additional analyses are conducted to evaluate the framework’s performance, including computational acceleration, sensitivity to specific parameters, and the influence of initial guesses. The proposed framework not only eliminates the dependency on initial guesses decided by designers in conventional topology optimization methods but also enables efficient execution of the metaheuristic search. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88492 |
DOI: | 10.6342/NTU202302448 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 機械工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf 此日期後於網路公開 2025-08-01 | 3.99 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。