LLDPE/EVA/ZnO纳米复合材料的制备与BP-Markov性能预测研究  

Preparation and prediction based on BP-Markov model of LLDPE/EVA/nano-ZnO composite material

作　　者：刘佳[1] 李顺祥[2] 贾树盛[3] 

机构地区：[1]广东工业大学材料与能源学院,广东广州510006 [2]广东工业大学自动化学院,广东广州510006 [3]吉林大学汽车材料教育部重点实验室,吉林长春130022

出　　处：《功能材料》2009年第11期1925-1928,共4页Journal of Functional Materials

基　　金：国家重点基础研究发展规划(973计划)资助项目(2004CB619301)

摘　　要：制备了LLDPE/EVA/ZnO纳米复合材料,并测试了其力学性能。结合BP神经网络和Markov链建立了BP-Markov模型,并利用此模型对LLDPE/EVA/ZnO纳米复合材料的多指标性能进行了预测。结果表明BP-Markov模型应用于聚合物/无机物纳米复合材料的多指标性能预测具有较高的精度和可靠性,Markov链解决了神经网络在做多指标性能预测时误差的随机性和波动性问题,充分的发挥了神经网络与Markov链预测模型的优点,为在数据有限且具有随机因素的情况下实验数据的分析提供了一种新的思路。LLDPE/EVA/ ZnO nanocomposite material were prepared, and the mechanical properties were tested. A BP-Markov model was constituted in the base of BP neutral network and Markov chain. And the multitarget properties of EVA/LLDPE/ZnO nanocomposite material were predicted by this BP-Markov model. It was showed that the method was effective and feasible in the use of the multi-target properties prediction of polymer/inorganic nanocomposites material. When the multi-target properties of polymer/inorganic nanocomposites material were predicted, the randomicity and fluctuation of the prediction errors were solved by the Markov chain. The advantages of BP neutral network and Markov chain were exerted sufficiently. That offered a kind of new ideas for analysis of experiment data, under the condition of limited data and stochastic factors.

关 键 词：聚合物/无机物纳米复合材料 正交实验 BP-Markov模型 性能预测 

分 类 号：TB332[一般工业技术—材料科学与工程]