【摘要】 应用人工神经网络研究了热固性酚醛树脂的合成反应条件与其玻纤复合材料力学性能之间的映射关系。建立了可用于定量预测玻纤增强酚醛树脂静弯曲强度的模型 ,并进行了实验验证。结果表明 ,当醛 /酚比取 1.3～ 1.5时 ,相对于 Mg(OH) 2 、Ba(OH) 2 和 Na OH而言 ,NH4OH催化所得酚醛树脂的复合材料在室温与 2 5 0℃均有较高的静弯曲强度。神经网络方法对聚合物基复合材料的反应 -性能定量关系研究显示出良好的应用前景。更多还原

【Abstract】 An approach of Artificial neural networks(ANNs) was made for studying the complicated nonlinear correlation between synthesis reaction conditions and mechanical properties of thermosetting phenolic resins (resoles) based composites. The quantitative prediction models for static flexural strengths of glass fiber reinforced resole composites were established by using multilayer neural networks with an improved back propagation algorithm. The results prove that composites reinforced by resoles, which were catalyzed by NH 4OH, instead of Mg(OH) 2, Ba(OH) 2 and NaOH have relatively higher static flexural strengths at both room temperature and 250 ℃, when the formaldehyde phenol molar ratio was between 1.3~1.5. With excellent predicting performance. ANN was proved a promising tool in the study of quantitative correlation between synthesis conditions and properties of polymer based composites.更多还原