摘要
将基础油剂、偏亲水乳化剂和偏亲油乳化剂按不同配比制备涤纶全拉伸丝(FDY)油剂以及油剂质量分数为1%的乳液,定量分析油剂及乳液的外观随基础油剂含量、偏亲水乳化剂含量的变化规律;建立以基础油剂含量及偏亲水乳化剂含量为输入、油剂及乳液的外观为输出的分类神经网络模型,用采集的不同配方下油剂及乳液的外观进行网络训练,用新配方下油剂及乳液的外观进行模型验证。结果表明:油剂及乳液的外观是关于基础油剂含量、偏亲水乳化剂含量的分段函数,在基础油剂含量确定时,随着偏亲水乳化剂含量增加,油剂及乳液的外观依次呈现出白色乳液、略带蓝光白色乳液、蓝白色半透明液体、透明液体和暗灰色液体,且油剂分层;分类神经网络模型经训练后,准确率达98.3%,模型拟合效果好,能准确预测新配方下油剂及乳液的外观,可以为判断新油剂配方下能否制备微乳液体系提供帮助。
A series of polyester fully-drawn yarn(FDY)finish and its corresponding emulsion containing 1%finish by mass fraction were prepared under different proportions of base oil,hydrophilic emulsifier and lipophilic emulsifier.The change law of the appearance of finish and emulsion with the contents of base oil and hydrophilic emulsifier was analyzed quantitatively.A classification neural network model was established with the contents of base oil and hydrophilic emulsifier as the input and the appearance of the finish and emulsion as the output.Network training was carried out with the appearance of finish and emulsion at different formulations,then the model validation was performed with the appearance of finish and emulsion at new formulations.The results showed that the appearance of finish and emulsion was a piecewise function of the contents of base oil and hydrophilic emulsifier;the appearance of finish and emulsion showed white emulsion,white emulsion with blue light,blue-white translucent liquid,transparent liquid and dark gray liquid and finish layering in turn with the increase of the content of hydrophilic emulsifier at a certain content of base oil;the accuracy of the classification neural network model was 98.3%and the model approximation(fitting)effect was good after training;and the model can accurately predict the appearance of finish and emulsion under new formulations,which help to evaluate whether the micro-emulsion system could be prepared under new finish formulation.
作者
郑征
杨以琳
马剑斌
徐锦龙
王松林
陈伟波
ZHENG Zheng;YANG Yilin;MA Jianbin;XU Jinlong;WANG Songlin;CHEN Weibo(Zhejiang Henglan Technology Co.,Ltd.,Hangzhou 311215;School of Chemical Engineering,Carnegie Mellon University,Pittsburgh PA15213,USA)
出处
《合成纤维工业》
CAS
2020年第5期16-21,共6页
China Synthetic Fiber Industry
关键词
油剂
乳液
微乳液
外观
分类神经网络模型
预测
finish
emulsion
micro-emulsion
appearance
classification neural network model
prediction
