應用冪次律型損失函數與灰關聯分析於多重品質輪磨過程最佳化分析

利用磨床加工來進行表面粗糙度及平坦度多重品質最佳化分析。實驗分為先前實驗、單一品質實驗及多重品質實驗。實驗設計採用田口直交表實驗分析，以損失函數冪次律型代替二次曲線，灰關聯分析將表面粗糙度及平坦度值結合，找出多重品質特性最具影響力的貢獻因子及最佳參數組合，並進行確認實驗與預測值相互比較。單一品質實驗結果顯示，單一表面粗糙度最佳製程參數為A1B3C1D1E3，確認實驗與預測值誤差百分比皆在2%左右。單一平坦度方面則得到兩組最佳化參數組合，經由冪次律損失函數分析得到最佳參數組合為A2B1C2D1E1，確認實驗與預測值之誤差百分比為14%左右。多重品質實驗結果顯示，得到最佳製程參數組合為A2B1C2D1E1，確認實驗與預測值之誤差百分比為2%到7%之間，證實多重品質特性最佳製程參數組合為可行性的加工製程參數組合。

關鍵字

輪磨；田口方法；灰關聯分析；表面粗糙度；平坦度

並列摘要

Optimization of surface roughness and flatness are using grinding process with multiple performance characteristics. There have advance, single and multiple performance experiment. Using the orthogonal array of Taguchi experimental analysis with experimental design, and then instead the curve of the second with power-law loss function. Combined the surface roughness and flatness used grey relational analysis to find contribution factor and optimal processing conditions of multiple performance characteristics, and then to compare experimental and predicted values. Experimental results show that with single quality, optimization of the surface roughness parameter is A1B3C1D1E3 determine experiment and predicted values error about two percent. The single quality of flatness were have two optimization parameters and then optimal processing conditions is A2B1C2D1E1 with power-law loss function, the determine experiment and predicted values error was about fourteen percent. Experimental results show that with multiple performance characteristics the optimal processing conditions is A2B1C2D1E1 the determine experiment and predicted values error were between the two and seven percent, the experiment is good to prove of optimal processing conditions by multiple performance characteristics analysis.