基于小波分析理论与BP神经网络的复合阻燃剂性能优选预测  

Structural Analysis and Optimization Performance Forecast of Flame Retardant Composite Based on Wavelet Analysis and Back-propagation Neural Network

作　　者：张庆丰[1,2] 陆彪[2] 

机构地区：[1]马鞍山职业技术学院电气工程系,安徽马鞍山243031 [2]东南大学能源与环境学院,江苏南京210096

出　　处：《稀土》2016年第4期29-34,共6页Chinese Rare Earths

基　　金：安徽省高校省级优秀青年人才基金重点项目(2013SQRL145ZD)

摘　　要：以正硅酸乙酯、磷酸二氢铵、尿素、季戊四醇、氯化镧制备复合阻燃剂。研究氯化镧添加量对复合阻燃剂结构的影响规律,通过傅里叶红外光谱分析氯化镧增效机理,研究结构参数与性能的复合阻燃剂性能优选预测模型,通过小波分析理论与BP神经网络拟合各结构参数与性能的关系。结果表明,氯化镧对复合阻燃剂具有增效效果;当氯化镧添加量0.033时,复合阻燃剂性能最优;性能优选预测模型的预测值与实测值吻合较好,其相对误差为-3.80%-3.28%。Flame retardant composite was prepared with tetraethyl orthosilicate, ammonium dihydrogen phosphate, urea, pentaerythritol and lanthanum chloride. Influence regulations of amount lanthanum chloride on structure of flame retardant composite was researched, and the synergistic mechanism of lanthanum chloride was analyzed by FTIR. Optimization perform- ance forecast model for flame retardant composite was estabilished, and the relationship of structure parameters and perform- ance was fitted by wavelet analysis and back - propagation neural network. The results show that lanthanum chloride has syner- gistic effect o flame retardant composite ; the best performance of flame retardant composite was obtained with 0. 033 lanthanum chlorid. Predicted data with optimization performance forecast model agree well with the experimental data, and the relative er- ror is -3.80% -3.28%.

关 键 词：氯化镧 小波分析理论 BP神经网络 傅里叶红外光谱 复合阻燃剂 性能预测模型 

分 类 号：TQ323.8[化学工程—合成树脂塑料工业]