Online defect prognostic model for textile manufacturing

Abstract

Defects in textile manufacturing process cause a significant waste of resources and affect the quality of product. This is challenging manufacturers to maintain growth and competitiveness in Industry 4.0. To solve the problem, models for predicting defects should be developed to assist shop floor operators. However, there is a lack of studies and models for solving the problem. Focusing on realistic needs, this study aims to develop online defect prognostic models for textile manufacturing. In particular, data from the manufacturing processes are collected in time series. Then, control charts are used to transform the collected data into region data of product. Based on these data, back-propagation neural networks are used for predicting defects at each stage. In addition, an experiment was conducted to validate the proposed approach. The results have shown the robustness and efficiency of the proposed model. This model can be implemented in practice to predict defects in advance that assists operators taking correct actions to prevent defect products and reduce waste.