Prediction of mechanical and thermal properties in bronze-filled polyamide 66 composites using artificial neural network
Prediction of mechanical and thermal properties in bronze-filled polyamide 66 composites using artificial neural network
复制标题
使用人工神经网络预测青铜填充聚酰胺 66 复合材料的机械和热性能
DOI：
10.1007/s00289-021-03751-5
复制
发表时间：
2021
期刊：
影响因子：
3.2
通讯作者：
M. Aliasghary
中科院分区：
文献类型：
--
作者：
Mahboube Mohamadi;S. Alavitabari;M. Aliasghary
在同向双螺杆挤出机中制备了低含量青铜粉（3、5和7wt%）增强聚酰胺66（PA 66）的微复合材料。机械性能，包括拉伸特性，耐冲击性，和海岸D硬度，进行了评价。结果表明，随着青铜粉用量的增加，复合材料的断裂伸长率和冲击强度降低，而硬度在青铜粉用量为7wt%时达到最大值（提高15%）。利用扫描电子显微镜（SEM）对复合材料的断口形貌进行了分析，探讨了复合材料的增韧机理.作为尺寸稳定性的良好指标的热膨胀系数通过应用热机械分析（TMA）来测量。通过人工神经网络（ANN）方法对实验测得的机械和热性能进行建模。该网络采用Levenberg-Marquardt反向传播（LMBP）在由5个神经元组成的单个隐层中进行训练。基于人工神经网络预测和经验结果之间的良好一致性，人工神经网络模型可以被认为是一个可靠的工具，估计和评估材料的合成和制造之前的性能。
Microcomposites based on polyamide 66 (PA66) reinforced with bronze powder in low contents (3, 5 and, 7 wt%) were prepared in a co-rotating twin-screw extruder. Mechanical performance, including tensile characteristics, impact resistance, and Shore D hardness, was evaluated. The results indicated that the elongation at break and impact strength decreased with the increase in bronze loading, while the hardness reached a maximum (15% enhancement) when using 7 wt% of bronze powder. Scanning electron microscopy (SEM) was utilized to analyze the fracture surface and study the toughening mechanisms. The thermal expansion coefficient, as a good indicator of dimensional stability, was measured by applying thermomechanical analysis (TMA). The experimentally measured mechanical and thermal properties were modeled by an artificial neural network (ANN) method. The network was trained by Levenberg–Marquardt back-propagation (LMBP) in a single hidden layer which is consist of five neurons. Based on the excellent consistency between the ANN predictions and empirical results, ANN models can be considered as a reliable tool to estimate and evaluate material properties before synthesis and manufacturing.