| 摘要: | 本研究對金屬成形製程進行了全面的統計分析與人工智能應用，重點在於自衝鉚接
（SPR）的預測模型建構，以及增材積層輔助投資精密鑄造的永續性績效評估。研究的第一部分，透過實驗觀察、基於QForm之模擬以及人工神經網絡（ANN）模型，探討了鋁與高強度鋼之間的自衝鉚接（SPR）製程，旨在預測接頭性能。結果顯示，ANN模型實現了高預測精度，其機械互鎖值的平均絕對百分比誤差（MAPE）為7.56%，能量吸收則為11.02%，突顯了其在優化異質材料SPR接頭方面的卓越能力。第二部分比較了傳統單階段脫蠟精密鑄造（方案A）、傳統雙階段射蠟製程（方案B）以及基於增材積層製造的方案（方案C）。方案C展現出最高的尺寸精度（偏差為0.02–0.32毫米；標準差為0.01–
0.08毫米），使其成為最適合小批量生產的方案。最後，本研究透過生命週期評估
（LCA）框架，評估了一項增材積層輔助脫蠟鑄造於不銹鋼歧管製造的實際應用案例。增材積層輔助脫蠟鑄造製程使成品硬度提升了7.59%，並顯著降低了生產週期時間、成本、能源消耗及環境影響，降幅分別達40.5%、42.0%、93.6%和93.6%。儘管具備這些優
勢，但考慮到大規模生產中與設備折舊相關的時間和成本限制，增材積層輔助脫蠟精密鑄造仍最適用於複雜、小批量的生產模式。總體而言，本研究證明，整合統計預測模型與增材積層製造策略能有效提升先進金屬成形與鑄造製程的精密度及永續性。
;This study presents a comprehensive statistical analysis of metal forming processes, focusing on predictive modeling for self-piercing riveting (SPR) and sustainability performance evaluation of Additive-Aided Investment Casting. As the first part, explores the self-piercing riveting (SPR) process between aluminum and high-strength steels through experimental observations, QForm- based simulations, and Artificial Neural Network (ANN) modeling aimed at predicting joint performance. The ANN model achieved high predictive accuracy, with Mean Absolute Percentage Errors (MAPE) of 7.56% for interlock and 11.02% for energy absorption, emphasizing its capability in optimizing multi-material SPR joints. The second part compares traditional investment casting with one stage (Scheme A), traditional process scheme with dual-stage (Scheme B), and additive manufacturing-based (Scheme C). Scheme C exhibited the highest dimensional accuracy (0.02–0.32 mm deviation; 0.01–0.08 mm standard deviation), making it most suitable for small-batch production. Finally, a real-case application of Additive-Aided Investment Casting for a stainless steel manifold was assessed through a Life Cycle Assessment (LCA) framework. The Additive-Aided Investment Casting process performs hardness by 7.59% and significantly reduced production cycle time, cost, energy consumption, and environmental impact by 40.5%, 42.0%, 93.6%, and 93.6%, respectively. Despite these advantages, Additive- Aided Investment Casting is best suited for complex, low-volume production due to time and cost constraints related to machine depreciation in mass manufacturing. Overall, this study demonstrates how integrating statistical predictive modeling and additive manufacturing strategies can enhance precision and sustainability in advanced metal forming and casting processes. |
