- 學位論文
利用類神經網路分析評估槍鑽深孔加工的切削力
AN ANALYTICAL EVALUATION OF THE CUTTING FORCES IN GUN DEEP DRILLING USING ARTIFICIAL NEURAL NETWORKS
摘要
為了增加槍鑽深孔加工之實用性與應用性，本研究針對槍鑽刀具幾何及加工參數，建立深孔加工之切削力學模式，並進一步探討各參數對切削阻抗的影響。首先，利用田口實驗方法進行實驗設計與配置，經由實際進行加工並量測主軸推力、扭力等切削阻抗。經由因子反應分析及資料變異分析。結果顯示，主軸進給與轉速為影響切削阻抗最重要因子，當進刀量愈大時，主軸推力與扭力愈高；且外切削角較內切削角因子效應顯著，當外切削角度愈大時主軸推力與扭力也愈高。本研究導入誘導式類神經網路系統建立切削力學模式。結果顯示，內、外傾角與鑽頂位置等刀具幾何並不影響主軸推力與扭力等切削阻抗，且由A.I.M.所得之力學模式相較於田口式線性模式能有較佳的預測結果，且誤差皆在10%以下。
並列摘要
In order to increase the practicability and application of the deep -hole drilling, a cutting force model aimed at cutting geometries and working parameters is developed to find out the influences of the cutting impedance. First of all, Taguchi Method is utilized for the design of experimental array. The thrust force and torque of the main spindle are practically measured and thereby investigated via the factor response analyses as well as analysis of variance. The result shows that the rotational speed and feed speed of the main spnidle are the most important factor influencing the cutting impedances, when the bigger revolutionary feed rate causes the larger thrust force and torque. Also, the factor effect of the outer cutting angle is more significant than that of the inner cutting angle, when the larger degree of the outer cutting angle has the higher cutting impedances. Furthermore, the artificial neural network system is applied to develop the cutting force model. The results reveal that the inner incline angle, the outer incline angle and the position of the drill top have insignificant factor effect on the cutting impedances. Compared with the Taguchi linear model, the cutting model developed by applied neural network system has the better performance of prediction, whose errors are all under 10%.