博碩士論文 110323049 完整後設資料紀錄

DC 欄位 語言
DC.contributor機械工程學系zh_TW
DC.creator曾煒策zh_TW
DC.creatorWei-Tse Tsengen_US
dc.date.accessioned2023-8-8T07:39:07Z
dc.date.available2023-8-8T07:39:07Z
dc.date.issued2023
dc.identifier.urihttp://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=110323049
dc.contributor.department機械工程學系zh_TW
DC.description國立中央大學zh_TW
DC.descriptionNational Central Universityen_US
dc.description.abstract本論文以不同模具設計與不同材料及厚度的鉚接為研究,進行了模擬驗證、鉚接與拉伸實驗,並以機器學習的人工神經網路(ANN)為輔助,以歸納出何種因素會對鉚接品質與機械性質造成影響。而詳細研究中我們使用了各種的雙相鋼(DP780、DP980、1180MS)和不同深度的模具以及板材的厚度進行實驗,由於傳統的點焊對於不同種類的金屬進行焊接是一件難以達成的目標,故此研究提出使用鉚接的方式,搭配有限元素法(FEM)對於鉚接時的材料流動及有效應力分布進行分析模擬驗證其金屬流動的過程,並提供品質表現指標(PM)與拉伸實驗後所得到的機械性質與能量吸收作為評估方式,最後將以上連結成一條龍式的製程,透過訓練模型,使未來無須透過實際實驗便能得出詳細的結果以避免不必要的成本浪費。 本研究的最佳組合為厚度1mm的1180MS搭配2mm的Al6061對應到2.25mm深度的下模具,其能夠承受的最大拉力為9.26kN,而透過計算所得出的能量吸收為36.02J,而ANN模型將各項組合的數據進行匯入後訓練出的成果包含幾何尺寸的互鎖尺寸(interlock)、殘餘厚度(remaining thickness)以及作為品質的PM值,還有機械性質相關的最大承受拉力和能量吸收後,可以對使用者輸入的材料、厚度、模具深度進行預測,而針對上面所提到的輸入透過計算所得到的平均絕對誤差(MAE,Mean Absolute Error)分別為:0.031mm、0.051mm、0.029、0.714kN和3.38J,對應到的平均絕對百分比誤差(MAPE,Mean Absolute Percentage Error)為8.53%、12.99%、16.8%、11.22%和13.96%。zh_TW
dc.description.abstractIn this study, we investigate the effects of different designs on the quality and strength of self-piercing riveting (SPR) in dissimilar metals. Specifically, we explore a wider range of dimensions that contribute to the overall strength of the joints. For this purpose, we prepared various dual-phase sheets of steel (DP780, DP980) and high-carbon steel (1180MS) to compare their strengths. Aluminum plates (Al6061) were also used consistently. We examine the shape and quality of the joints after riveting, and verify the trends of the important indicator, interlock, through simulations. Additionally, shear tests are conducted on the same samples to confirm the physical strength properties. Finally, the primary contribution of this study is to establish a correlation between the end-to-end process chain quality and shear test evaluations for self-piercing rivets of dissimilar metals, using Performance Metrics (PM) and mechanical properties, as well as energy absorption, as validation indicators. This study determined that the optimal combination is a die with a depth of 2.25mm, paired with a thickness of 1mm of 1180MS and 2mm of Al6061, capable of withstanding a maximum tensile force of 9.26kN and absorbing a maximum energy of 36.02 (J). This research employed an Artificial Neural Network (ANN) as a model to predict the quality and mechanical properties of riveting. The predicted outputs include interlock, remaining thickness, PM value, Maximum force, and energy absorption, serving as five reference indicators. The mean absolute errors (MAE) between the experimental and prediction results for the interlock, remaining thickness, PM value, maximum force and energy absorption reached 0.031mm, 0.051mm, 0.029, 0.714kN and 3.38J respectively, and the corresponding mean absolute percentage errors (MAPE) were 8.53 %, 12.99 %, 16.8 %, 11.22% and 13.96%. These results indicate a high level of prediction accuracy. Through validation against actual results, the study discusses the interrelationships between different riveting parameters and joint quality as well as mechanical properties.en_US
DC.subject自沖鉚釘zh_TW
DC.subject異質材zh_TW
DC.subject拉伸試驗zh_TW
DC.subject端到端製程zh_TW
DC.subjectSelf-piercing riveten_US
DC.subjectDissimilar metalsen_US
DC.subjectShear testen_US
DC.subjectEnd-to-end process chain qualityen_US
DC.title機器學習輔助自沖鉚釘於鋁合金6061和雙相鋼(DP780、DP980、1180MS)端到端製程中之品質評估方法(鉚接品質和機械性質)zh_TW
dc.language.isozh-TWzh-TW
DC.titleMachine learning assisted evaluation methodology of selfpiercing rivets end-to-end process chain quality (joint quality and mechanical properties) for aluminum alloy 6061 and dual-phase steels (DP780,980,1180)en_US
DC.type博碩士論文zh_TW
DC.typethesisen_US
DC.publisherNational Central Universityen_US

若有論文相關問題,請聯絡國立中央大學圖書館推廣服務組 TEL:(03)422-7151轉57407,或E-mail聯絡  - 隱私權政策聲明