應用類神經網路於製程平均值與變異性變化之偵測及參數之估計

A Neural Network Approach for Process Mean and Variance Changes Detection and Parameter Estimation

楊中孚(CHUNG-FU YANG)

指導教授 ： 鄭春生博士

元智大學/工程學院/工業工程與管理學系/碩士(2001年)

摘要

管制圖是被廣泛應用來監控製程狀態的重要工具。管制圖的應用可以了解製程是否在管制內或發生異常。近年來，類神經網路 (artificial neural networks) 藉由仿效人類思考及處理資訊的方式，已成功應用於管制圖的分析上。類神經網路透過製程中所收集之樣本數據，可用來分析製程之狀態。
本研究發展一個以類神經網路為基之兩階段式偵測系統，用來偵測製程異常類型及估計重要參數。當第一階段之類神經網路偵測出異常類型後，隨即進入第二階段之類神經網路來估計其參數變化量。本研究所建立之系統是以辨識率及絕對平均誤差百分比來評估效益。經由模擬方法分析，證明本研究所提出之類神經網路可以有效地辨認異常類型及估計參數之變化量，其成效優於傳統之蕭華特管制圖。

關鍵字

蕭華特管制圖 ； 類神經網路 ； 辨識率

並列摘要

Control charts are widely used for controlling industrial processes. They are useful in determining whether a process is behaving as intended or if there are some unnatural causes of variation. Recently, artificial neural networks that attempt to emulate the massively parallel and distributed processing of the brain are being examined for use in control chart analysis. Artificial neural networks can be used to analyze process status from input of control chart samples.
This paper describes a neural network based approach to monitor the process mean and variance changes and to predict change magnitudes. A two-stage approach is proposed in this research. At the first stage, a back-propagation neural network will be used to detect process mean and/or variance shifts. At the second stage, a second back- propagation neural network is used to estimate the magnitude of shifts as soon as the first neural network detects a shift in the mean and/or variance. The performances of neural networks were evaluated by estimating the percentage of correct classification and mean absolute percent errors using simulation. The simulation results indicate that the proposed neural networks outperform traditional Shewhart control charts.

並列關鍵字

Shewhart control charts ； artificial neural networks ； recognition rate

參考文獻

詳見論文