The Journal of The Textile Institute ( IF 1.7 ) Pub Date : 2020-09-14 , DOI: 10.1080/00405000.2020.1812921 Assad Farooq 1 , Farida Irshad 1 , Rizwan Azeemi 2 , Muhammad Nadeem 2 , Usama Nasir 1
Abstract
Finishes are applied to improve the look, performance and feel of the fabrics. Crease recovery finishes form a three-dimensional crosslinking network on the surface of the cotton knitted fabric to control its dimensions. However, application of the crease recovery finishes induces the shade change in the dyed fabrics. This paper presents the phenomenon of shade change for different colors and shade percentages and use of artificial intelligence-based prediction system to foresee the behavior of shade after finish application. The individual neural networks were trained for the prediction of color of the finished samples, which are delta color coordinates values (△L, △a, △b, △c & △h). The input variables, i.e. reflectance values (Visible ranges 400–700 nm) of dyed samples, color, shade percentage and finish concentration were used to train the networks. The trained neural networks were validated through ‘cross validation’ and ‘hold out’ techniques. The shade prediction model was developed by combining the individually trained artificial neural networks and the developed model can predict the shade change with more than 90% accuracy. This will help the dyers to predict shade change prior to dyeing & finishing and they will adjust their recipe accordingly, which can ultimately reduce the rework and reprocessing in the textile wet processing industries.
中文翻译:
使用人工神经网络开发阴影预测系统以量化折痕恢复整理应用后的阴影变化
摘要
整理剂用于改善织物的外观、性能和手感。折痕恢复整理剂在棉针织物表面形成三维交联网络,以控制其尺寸。然而,折痕恢复整理剂的应用会引起染色织物的色调变化。本文介绍了不同颜色和色调百分比的色调变化现象,并使用基于人工智能的预测系统来预测涂饰后的色调行为。单独的神经网络被训练用于预测成品样本的颜色,这些颜色是颜色坐标值(△L、△a、△b、△c 和△h)。输入变量,即染色样品的反射率值(可见光范围 400–700 nm)、颜色、阴影百分比和饰面浓度用于训练网络。训练有素的神经网络通过“交叉验证”和“保持”技术进行验证。阴影预测模型是通过结合单独训练的人工神经网络开发的,开发的模型可以以超过 90% 的准确度预测阴影变化。这将有助于染工在染整前预测色调变化,并相应地调整配方,最终减少纺织品湿加工行业的返工和再加工。阴影预测模型是通过结合单独训练的人工神经网络开发的,开发的模型可以以超过 90% 的准确度预测阴影变化。这将有助于染工在染整前预测色调变化,并相应地调整配方,最终减少纺织品湿加工行业的返工和再加工。阴影预测模型是通过结合单独训练的人工神经网络开发的,开发的模型可以以超过 90% 的准确度预测阴影变化。这将有助于染工在染整前预测色调变化,并相应地调整配方,最终减少纺织品湿加工行业的返工和再加工。