麻灰纱中色纤维混合模型运用

针对目前大部分色纺企业仍然依靠有经验的配色人员进行人工配色,存在配色效率低、配色精度差等问题,提出运用反向传播(BP)神经网络的方法对色纺纱的黑白纤维混合配色进行预测,并与使用Datacolor MATCH系统模拟染料配色方法和基于颜色混合模型中的Kubelka-Munk双常数理论的配色方法对黑白纤维混合配色的结果进行对比。结果表明:上述3种方法均可对麻灰纱的黑白纤维混合配色进行有效的预测,配方的相对误差基本控制在7.36%之内,且配方样品与标准样品的色差小于1;比较而言,3种黑白纤维混合配色的预测模型中,基于BP神经网络的配色方法适用性及精度最佳,配方的相对误差最高,为3.08%。

Application of colored fiber mixed models in gray spun yarn

At present, most of color spinning enterprises still rely on experienced color matching persons for color matching, and some problems such as low color matching efficiency and poor accuracy still exist in the production. In order to solve these problems, back propogation( BP) neural network method was proposed to predict black and white fiber color matching in comparison with the prediction results using the Datacolor MATCH system simulation method and color mixing based model Kubelka-Munk two-constant theory. The above-mentioned three methods were all determined to be effective in predicting color mixing of black fiber and white fiber in gray spun yarns. The relative errors were controlled within 736%, and the color differences between formula and standard samples were less than 1 . It is found that the matching method based on BP neural network shows the optimal applicability and accuracy, and the relative error is below 308%.