Logo image
望小品質特性非常態製程能力分析之系統化研究
Dissertation

望小品質特性非常態製程能力分析之系統化研究

劉培熙
Doctor of Philosophy (PHD), 國立清華大學, 工業工程與工程管理學系
2003

Abstract

製程能力分析 非常態製程 望小品質特性 近似分配函數 當量製程能力指標 Process Capability Analysis Non-normal processes Smaller-the-Better Quality Approximation Distribution Function “Equivalent"Cpu
ABSTRACT Process capability analysis has been widely used and applied in industrial production processes. The objective of a process capability analysis is to estimate, monitor, and furthermore may reduce variability in industrial production processes. Additionally, process capability analysis provides a common standard of product quality for suppliers and customers. The most popular way to assess process capability is to use histograms and process capability indices (PCIs). In practice, there are many non-normal processes existed in industrial production processes, especially in the precision industrial processes, so that the use of PCIs based on the normality assumption can be misleading in many cases. In this research, we propose a new model to evaluate the non-normal process capability with smaller-the-better quality characteristic in order to achieve three goals: (1) to reflect the process status more realistically and to identify the quality characteristics of the non-normal process correctly, (2) to define, interpret and use PCIs adequately, and (3) to evaluate the non-normal process capability properly. This new model includes three functional modules. The data collection and clustering module collects and analyzes the data of process quality characteristics to find its own data types, and then separate the data appropriately into some data segments. The distribution-fitted module treats each data segment by goodness of fit method to find their approximation distribution function, and then calculates the process control limits, percentiles and the upper bound of defective rate. Finally, the PCIs evaluating module defines and calculates individual PCI and aggregate PCI of the process ( particularly for Cpu and “Equivalent”Cpu ) based on yield information. After experimenting with some examples for processes of smaller-the-better quality characteristic, It shows that this approach could avoid besetments and mistakes which may be exists when using traditional methods. Keywords: Process Capability Analysis, Non-normal processes, Smaller-the-Better Quality, Approximation Distribution Function, “Equivalent”Cpu

Metrics

1 Record Views

Details

Logo image