基于主成分分析与BP神经网络预测织机效率

Loom Efficiency Prediction Based on Principal Component Analysis and BP Neural Network

为降低企业生产成本,通过对生产工艺参数进行调整,提出一种织机效率预测模型。该模型将主成分分析与BP神经网络结合,先用主成分分析法对影响织机效率的众多因素进行预处理,降低原变量的维数,消除原变量之间的相关性。然后再将经过预处理的主成分作为神经网络的输入,这样不仅简化网络结构,还能提高网络稳定性。经过仿真,结果表明,PCA-BP比BP神经网络相关系数高;十万纬经停仿真,PCA-BP比BP神经网络预测误差减小了11.28%;织机效率仿真,PCA-BP比BP神经网络预测误差减小了64.92%。

Based on the basic properties of chitin,the morphology structure and infrared spectrum were tested at first.Then the strength,evenness,yarn twist,yarn moisture content,yarn heat resistance and yarn acid resistance were tested which consists of chitin and cotton(20∶80).The results show that the cross section of chitin fiber is a circular or nearly circular,smooth vertical and a higher degree of orientation,and it contains amide group,carbonyl group,hydroxyl group and other functional groups;the moisture regain and moisture content of blended yarn are 7.52%and 7.23%respectively;And the yarn twist unevenness is uniform,moderate,low rate of yarn uniform.The blended yarn is suitable for weak acid,weak alkaline and neutral conditions of processing,the temperature is not too high.The chitin fiber knitted fabrics has excellent wearing properties and good healthcare function.