現代化的產品及其生產過程日趨複雜，一個產品或製程的品質往往取決於?個或?個以上彼此相關的品質特性，所以?用單變?管制圖做個別的監控容?產生誤判的機會，故採用多變?管制圖?同時監控?個產品品質特性就顯得相當需要而且重要?。主成分分析法屬於多變?投影法?用統計與線性代?的原?，將?多可能具相關性變?空間，投影至較低維?且完全線性獨?的潛存變?空間，以擷取大?變?產生龐大?據隱含的特徵訊息。本?文主要?用多變?統計的主成份分析，針對製程所產生的變?，進?多變量分析模型的建?，建構錯誤診斷系統，即時掌握整體製程變?的變??態，並加以驗證。本研究在高爾夫球頭製程中，以多變量分析方式探討生產參數的影響性，以資?矩陣大小為99(?測?目) × 3 (變??目)，建?主成份模型，再針對變?變化建?一個錯誤診斷系統，其結果利用T?管制圖能正確找出裂縫發生之錯誤球頭，並利用MYT法找出錯誤參數。 The manufacturing process today is very complicated for the quality of products is usually decided by two or more related characteristics. Therefore it is necessary and important to develop a multi-variables control charts on monitoring several characteristics of product.Principal Component Analysis (PCA) is one of the multivariate projection methods, using principles of statistics and linear algebra to project variable spaces of samples in possible to Latent spaces which have low dimensions and linear independent. We set variables and transfer to eigenvector space, which corresponding eigenvalues mean statistic weighting, these space configured by so-call principle components. Through building and verifying models in raw data, it could determine fault process occurrence.PCA is applied for a golf head process to investigate the impact of process parameters. Then, Hotelling’s T2 control charts are utilized to detect which golf head is defective and MYT method is implemented to reveal which parameter is the key factor for the defect.
