基于BP神经网络的熔喷非织造布工艺参数优化  

Optimizing technological parameters of manufacturing meltblown nonwovens by BP neural networks

作　　者：吴雄华[1] 刘亚[1,2] 

机构地区：[1]天津工业大学理学院,天津300160 [2]天津工业大学纺织学院,天津300160

出　　处：《纺织学报》2011年第1期51-54,共4页Journal of Textile Research

摘　　要：为得到熔喷非织造布的最优工艺参数,用实验数据中每组条件的前4个数据作为训练样本,建立了BP神经网络预测模型,并用其余数据作为检验样本对模型进行验证。结果表明:BP神经网络预测过滤效率和透气量的平均值时相对误差均小于9%,模型相关系数都接近于1,因此BP神经网络的拟合效果非常理想。在此基础上通过计算机模拟方法得到最优工艺参数,即温度、接收距离和面密度分别为281℃、18 cm和469 g/m2时,过滤效率可达到94%,透气量在400 L/(m2.s)以上。In order to get the optimal technological parameters of manufacturing mehblown nonwovens, the BP neural networks prediction model was established by using the first four data of every group as training input and output vectors in the experiment, while others as testing vectors to check the model. The results showed that the relative errors were less than 9% when the BP neural networks model was applied to predict the average values of filtration efficiency and permeability, and the correlation coefficients were both close to 1. It showed that the fitting effect of the BP networks model was very ideal. On this basis, the optimal technological parameters were obtained by computer simulation, that is, temperature of 218 ℃ , DCD of 18 cm, and surface density of 469 g/m2. Under these conditions, the filtration efficiency reached 94% and the air permeability was above 400 L/（m^2.s）

关 键 词：BP神经网络 熔喷非织造布 最优工艺参数 计算机模拟 

分 类 号：TS101.2[轻工技术与工程—纺织工程]