利用機器學習方法預測紡織準備段操作參數

Abstract

在工業4.0發展中，以資料科學的角度整合自動化實務和管理經驗，藉以改善製程品質與生產效能，是製造業轉型中重要的一環。本研究採用機器學習方式，對於不同紡織製品預測紡織梭織製程準備段的十五個機台操作參數，希望改善目前紡織產業大多需要依靠專家經驗進行相關判斷的缺點。本研究針對不同織品的紡織原料數據，利用特徵選取的方法找出關鍵特徵，並使用四種不同的迴歸分析模型對於紡織製程準備段參數進行預測，分析比對其正確率。交叉驗證實驗結果顯示，採用Linear Regression演算法的機器學習模型在梭織製程準備段的參數預測中表現最為優異，其均方誤差可縮小至0.056%。

Today, in an Industry 4.0 factory, improving quality of product and efficiency of manufacturing process are important goals in the textile industry. The utilization of advance-prediction tools, so that data can be systematically processed into information and thereby make more informed decisions. To achieve these goals, this study integrates automatic production process and management experience from the perspective of data science and provides a method using 4 different machine learning models to predict the machine operating parameters of the preparation section in fabric manufacturing. The purpose of this study is to intellectualize the textile manufacturing process in order to avoid the tech skills gap and also higher the efficiency of the textile manufacturing process. Based on 10-fold cross-validation, experimental results show that the proposed method provides good performance when comparing with previous stochastic methods, and the best regression model for predicting preparation section parameters can reduce the mean square error (MSE) to 0.056%