Application of Artificial Neural Network (ANN) to Hot Extrusion of AZ61 Magnesium Alloy Structural Parts

類神經網路應用於AZ61鎂合金結構件熱間擠製加工之探討

向四海(Su-Hai Hsiang) ； 林益瑋(Yi-Wei Lin) ； 李倉誠(Tsang-Cheng Lee)

《中國機械工程學刊》 31卷1期 (2010/02) Pp. 67-73

https://doi.org/10.29979/JCSME.201002.0008

摘要

鎂合金擠製加工過程中，許多製程參數皆會影響成品之機械性質，其中以胚料加熱溫度與擠製速度影響最大。本文將使用類神經網路方法進行AZ61鎂合金結構件熱間擠製加工後之抗拉強度之分析，以較少數之鎂合金結構件熱間擠製實驗資料，及所得之抗拉強度資料加入本研究室所建立之ANN資料庫進行學習，分析胚料加熱溫度處於300~340℃之間與擠製速度在1~3mm/s之間時，其結構件抗拉強度之變化情形，並進行驗證實驗探討ANN分析結果之準確性。最後，將結構件成品進行斷面硬度試驗，以了解擠製後成品之硬度分佈狀況，且與AZ31鎂合金及A6061鋁合金結構件之機械性質比較，探討三種材料之差異。

並列摘要

Many process parameters affect the mechanical properties of products of the extrusion of magnesium alloy. Among them, the heating temperature of billet and the extrusion speed have the greatest effect. Hence, in this study, the artificial neural network (ANN) method is applied to analyze the tensile strength of structural parts of the AZ61 magnesium alloy formed by hot extrusion. A few experimental data on the structural parts of hot-extruded magnesium alloy and acquired tensile strength data are added to the ANN database, established by our laboratory for learning, and the change in the tensile strength of the structural parts upon heating of the temperature of the billet to 300℃~340℃ with extrusion speeds of 1 mm/s~3 mm/s is analyzed. The study also proves the accuracy of the experimental ANN analytical results. Finally, a sectional hardness test is performed on the product to understand the hardness distribution of the extruded product. The mechanical properties of AZ61 magnesium alloy are compared with those of AZ31 magnesium alloy and A6061 aluminum alloy to understand the difference among these three materials.

並列關鍵字

Magnesium alloy structural parts ； ANN ； hot extrusion ； mechanical properties