摘要
针对纺织品热湿阻估计问题,采用纤维种类、面密度、厚度、透气性和回潮率5个影响因素,基于线性回归模型、BP人工神经网络模型,建立组合神经网络模型,分别对织物热湿阻值进行了实验估计。结果表明,线性回归模型对织物热阻和湿阻的估计精度平均值分别为7.97%和6.69%;BP人工神经网络模型的估计精度平均值分别为4.23%和4.72%;组合神经网络模型的估计精度分别为1.31%和1.97%,估计精度高于其他两种模型。
Linear regression model, BP neural network model and combination model were put forward to estimate thermal resistance and moisture resistance of fabric based on variety of fiber and cover density, thickness, permeability and moisture regain. The data from practice showed that the average estimation precision for thermal resistance and moisture resistance of fabric using linear regression model partly was 7.97% and 6.69%. The average estimation precision by BP neural network model reached to 4.23% for thermal resistance and 4.72% for moisture resistance. However, the combination model showed the best estimation precision, which was 1.31% for thermal resistance and 1.97% for moisture resistance. It afforded a convenience and effective way to do research on thermal resistance and moisture resistance of fabric.
出处
《丝绸》
CAS
北大核心
2008年第4期40-42,共3页
Journal of Silk
基金
河南省教育厅自然科学基金资助项目(2006110018)
关键词
纺织品热湿阻
舒适性
线性回归模型
BP神经网络
组合神经网络模型
Thermal & moisture resistance of fabric
Comfort
Linear regression model
BP neural network
Combination neural network model