一種結合循序向後選擇法與回歸樹分析的瑕疵肇因關鍵因子擷取方法與系統－以紡織製程為例;A Method and System for Extracting Key Factors Causing Defects Using Combining Sequential Backward Selection Method and Regression Tree Analysis: Taking Textile Manufacturing Process As an Example

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/84109

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/84109

Title:	一種結合循序向後選擇法與回歸樹分析的瑕疵肇因關鍵因子擷取方法與系統－以紡織製程為例;A Method and System for Extracting Key Factors Causing Defects Using Combining Sequential Backward Selection Method and Regression Tree Analysis: Taking Textile Manufacturing Process As an Example
Authors:	陳奕廷;Chen, Alvin
Contributors:	資訊工程學系
Keywords:	紡織製程;循序向後選擇法;回歸樹分析;最佳化參數;統計檢定;Textile process;Sequential backward selection;Regression tree analysis;Optimization parameters;Statistical verification
Date:	2020-08-13
Issue Date:	2020-09-02 18:05:36 (UTC+8)
Publisher:	國立中央大學
Abstract:	近幾年在紡織業中掀起短鏈革命的熱潮，在成品的交期越來越短的情況下，為了維持競爭優勢，減少紡織製程中產生的瑕疵，是一個勢在必行的問題。然而在紡織製程中會造成多種不同的瑕疵，每種瑕疵造成的原因也不盡相同，所以如何找出關鍵因子，並提出有效的最佳化參數設定，使其降低瑕疵發生的方法必須深入的探討。本研究根據紡織製程的資料集對個別的瑕疵種類使用循序向後選擇法找出影響瑕疵的關鍵因子，接著建立回歸樹模型找出具有較多低瑕疵數的葉節點，分析該節點具備的胚布性質與機台參數建立規則，最後使用統計檢定驗證規則是否有效降低瑕疵數。最後透過實驗發現理論上最多能夠為企業帶來39%的效益。;In recent years, there has been a short-chain revolution in the textile industry. With the delivery time of finished products getting shorter and shorter, in order to maintain a competitive advantage and reduce defects in the textile process, it is an imperative problem. However, a variety of different defects will be caused in the textile manufacturing process, and the cause of each type of defect is not the same. Therefore, how to find out the key features and propose effective optimization parameter settings to reduce the number of defects must be discussed in depth.This research uses the sequential backward selection method to find out the key features affecting the defects based on the data set of the textile process, and then establishes the regression tree model to find the leaf nodes with more low defects, and analyzes the fabric of the node. Establish rules for properties and machine parameters, and finally use statistical verification to verify whether the rules are effective in reducing the number of defects. Finally, through experiments, it is found that theoretically, it can bring 39% benefits to the enterprise at most.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	144	View/Open

社群 sharing

Loading...