十字沖頭擠鍛製程之最佳化分析與預測

Optimization Analysis and Prediction of Extrusion-Forging Processes for Philips Punch

許源泉(Yuan-Chuan Hsu) ； 林恆勝(Heng-Sheng Lin) ； 彭御芳(Yu-Fang Peng)

《鍛造》 19卷2期 (2010/06) Pp. 37-47

https://doi.org/10.29791/Forging.201006.0006

摘要

十字沖頭在進行冷擠鍛時，因製程參數控制不佳，易造成成形力過大或充填不足而導致模具應力增大與不良品增加等影響，致使生產成本增加與生產效率降低。現今實際生產上，製程參數皆由技術人員憑藉個人經驗或不斷試誤而得最佳製程參數，這將耗費許多時間及成本，所建立之資料亦不盡詳。故本研究提出了影響十字沖頭油壓擠鍛的成形力與充填品質的分析，先利用擠鍛實驗證實其結果與有限元素分析所得之結果之差異性；再以DEFORM-3D與田口法直交表之參數設計來簡化模擬分析次數，進行模具幾何最佳因子水準組合分析，取其中四個因子進行直交表之分析規劃，以尋求取其最佳製程參數；最後利用類神經網路建構十字沖頭擠鍛製程參數對成形負荷影響與充填品質之預測模式，以供業界對沖頭擠鍛的開發與設計作其參考依據。

關鍵字

十字沖頭 ； 擠鍛 ； 有限元素法 ； 田口法 ； 類神經網路

並列摘要

In cold extrusion-forging for making Philips punch, if the process parameters are not carefully managed, the die stress tends to be high due to excess forming load, and the ratio of failed products also increases with insufficient cavity filling, leading to higher production cost and lower production efficiency. In recent manufacturing industry, process parameters are usually obtained through the technicians' experiences or try-and-error methods, causing the waste of time and cost, and the technical information is also not fully established. In the current study, the forming load and filling quality in extrusion-forging for Philips punch has been analyzed. Firstly, the discrepancy of results between finite element analysis and extrusion-forging experiments was verified. Then, to obtain the optimal process's parameters and reduce the simulation quantity, the DEFORM-3D and Taguchi method were combined, using the orthogonal array of four factors, to analyze the optimal factor levels of die geometries. Finally, the neural network was employed to construct the prediction model presenting the effect of process parameters on forming loads and filling quality in Philips punch extrusion-forging. The result and approach obtained from this study would be beneficial to industries by providing the reference of the development and design of Philips punch extrusion-forging process.

並列關鍵字

Philips Punch ； Extrusion-Forging ； Finite element analysis ； Taguchi method ； Abductive network