應用 Minitab 於田口品質工程之數據重現與驗證：以血糖測試片射出成型論文為例

Abstract

中文摘要 塑膠射出成型在精密醫療檢測元件（如血糖驗測試片）的製造中扮演關鍵角色，然而，傳統試模階段往往耗費大量成本。田口方法與品質工程提供了一套能有效降低產品變異、實現製程穩健性的參數設計方法。本報告旨在透過實際演練，應用 Minitab 統計軟體對已發表之文獻《以田口方法導入一模多穴塑膠射出成型品質優化之實驗與數值模擬分析》進行實驗數據的重現與驗證。 研究過程中，本組成員引用 L18 直交表之 CAE 模流模擬與現場射出實驗數據，針對產品寬度、兩孔間距與上孔孔徑等關鍵尺寸之望目特性重新計算訊號雜訊比（S/N 比），並執行變異數分析。研究結果顯示，在「產品寬度」的實驗分析中，成功推導出與原論文一致的優化參數趨勢；然而，在深入驗證「兩孔間距」之現場原始數據時，Minitab 統計軟體系統判定其標準差趨近於 0，出現了不合理極端統計現象。經探討，真實多穴射出產線必然會受到環境與機遇原因的隨機雜訊干擾，數據理應具備自然的變異分散性。本報告推論，原始文獻之部分數據可能經過平滑化或以理想模擬輸出，而忽略隨機變異干擾因素；此外，針對「上孔孔徑」的變異數分析驗證中，本報告客觀指出僅有『料溫』達到統計顯著水準，修正了原文獻中對多項因子皆具顯著性判定。 本報告不僅完整演練田口方法實務操作流程，更透過客觀 Minitab 統計軟體運算，展現對學術文獻數據合理性批判性思考，期能為品質工程的學習與實務驗證提供具價值參考。

Abstract Plastic injection molding is critical for manufacturing precision medical components like blood glucose test strips, yet traditional mold tryouts are highly costly. Taguchi quality engineering provides a parameter design method to effectively reduce variation and achieve process robustness. This report uses Minitab statistical software to reproduce and validate experimental data from the published thesis, Experimental and Numerical Analysis on Quality Optimization of Plastic Injection of a Multi-Cavity Mold with Taguchi Method. Using the L18 orthogonal array's CAE simulation and on-site experimental data, we recalculated the signal-to-noise (S/N) ratios (nominal-the-best) and performed Analysis of Variance (ANOVA) for key dimensions: product width, distance between two holes, and upper hole diameter. While the optimized parameter trend for "product width" matched the original thesis, the raw on-site data for the "distance between two holes" exhibited an unreasonable statistical anomaly with a standard deviation approaching zero. Since real-world multi-cavity production inherently involves random noise, we infer that the original data was artificially smoothed or replaced by ideal simulation outputs. Furthermore, our ANOVA for the "upper hole diameter" objectively indicated that only the 'melt temperature' achieved statistical significance, correcting the original literature's determination of multiple significant factors. This report not only demonstrates the practical application of the Taguchi method but also showcases critical thinking regarding data validity through objective Minitab calculations, serving as a valuable reference for quality engineering.