基于均匀设计的神经网络在橡胶配方研究中的应用  

Application of the Artifical Neural Network Based on the Uniform Design in the Study of Rubber Formulation

作　　者：刘小青[1] 张振秀[1] 刘林国[1] 辛振祥[1] 

机构地区：[1]青岛科技大学橡塑材料与工程教育部重点实验室,山东青岛266042

出　　处：《青岛科技大学学报（自然科学版）》2011年第4期384-389,共6页Journal of Qingdao University of Science and Technology:Natural Science Edition

基　　金：青岛科技大学科研启动基金项目(0022446)

摘　　要：采用均匀设计方法设计了天然橡胶硫化体系配方,运用BP神经网络和遗传算法结合对实验数据进行了数据处理,用较少的实验次数成功建立了因素与性能之间的定量关系。利用训练好的神经网络对硫化配方进行优选,并对优选的配方方案进行试验验证,预测结果与实验数据比较吻合,并且得到当硫化配方为基本配方(均为质量份)天然橡胶100,炭黑N375 22.5,炭黑N660 22.5,增黏树脂4,硬脂酸3,氧化锌3.5,防老剂1.7时,硫磺1份,促进剂1.2份时,正硫化时间(TC90)较短,焦烧时间(TS2)较长,此配方是最优硫化配方。The uniform design method was employed in this paper to design the formulations of natural rubber vulcanization system,and the experimental data processing was adopted a hybrid artificial neural network（ANN）-genetic algorithm（GA） approach.The quantitative relations between the performance and influencing factors were gained in a few trials,and the the completed neural network was used to optimize the formulation of rubber.The predicted results was tested by experiments,and the results were good agreement with the predicted values.Besides,when the sulphr and accelerator loadings were 1 phr and 1.2 phr separately,the optimal curing formula,with shorter vulcanization time（TC90） and longer scorch time（TS2）,was obtained.

关 键 词：均匀设计 神经网络 遗传算法 硫化体系 配方优化 

分 类 号：TQ336.8[化学工程—橡胶工业]