模糊邏輯於品質管制圖診斷之應用

Abstract

管制圖為線上品質改善的重要工具之一。完整的管制圖作業程序應包管制 圖的解析、繪點、製程異常的追查與改善對策的發展、執行與標準化。但 目前的品質管制系統對製程管制的輔助大多局限於資料的儲存、統計分析 、管制圖的繪表等工作，對製程超出管制狀態的判別及製程問題原因的診 斷，仍有賴現場品管人員負責。本研究運用模糊邏輯的概念。提出一套輔 助管制圖的判讀與診斷異常原因的方法，並以一精密鑄造工廠之製程管制 作業為範例，發展一套即時回饋(real-time feedback)的品質管制圖診斷 系統的雛形(prototype )，提供X-bar管制圖的繪製、判讀及解析異常發 生原因的功能。 Control chart is a principal tool of on-line quality improvement. The implementation of process control should include analyzing the control chart pattern, diagnosing the troubles, formulating the adjust actions, and taking the adjust actions. Unfortunately, so far the decision support system for process control is limited to data storage, statistical analysis, control chart operation, etc. Therefore, the judgement and diagnosis for problems still depend on the experts, i.e. quality staffs. Using the concept of fuzzy logic, this study proposed a method to support the analysis of control chart pattern and diagnosis for the process problems. A prototype of real-time feedback control chart diagnosis system is developed for a casting plant. It provides the function of plotting, analyzing and diagnosing the X-bar chart.