论文部分内容阅读
以EVA(乙烯-醋酸乙烯酯)和淀粉质量比、甘油含量、NaHCO3含量为3个输入量,以拉伸强度和回弹率为输出量,建立3层BP(back propagation)神经网络,并将淀粉挤出发泡的正交实验结果作为样本对其进行训练,用以预测淀粉发泡材料的性能。研究结果证明,该BP神经网络能准确预测淀粉发泡材料的性能;同时发现,随着甘油含量的增加,淀粉发泡材料的回弹率逐渐增加,而拉伸强度则逐渐减小;NaHCO3发泡剂的质量分数为3%时,淀粉发泡材料的拉伸强度最小。研究结果将为提高生物质发泡材料的性能以及扩展其使用范围提供信息。
A three-layer BP (back propagation) neural network was established with three inputs of EVA (ethylene-vinyl acetate) and starch mass ratio, glycerol content and NaHCO3 content and the output of tensile strength and rebound rate. Starch extrusion foaming orthogonal experimental results as a sample to be trained to predict the performance of starch foam material. The results show that the BP neural network can predict the properties of starch foams accurately. At the same time, the resilience of starch foams increases with the increase of glycerol content, while the tensile strength decreases gradually. With the increase of glycerol content, When the mass fraction of the foaming agent is 3%, the tensile strength of the starch foaming material is the smallest. The findings will provide information on improving the performance of biomass foams and extending their use.