论文标题
预测人工神经网络中冻结铸造形成的孔隙率
Predicting the Porosity Formed in Freeze Casting by Artificial Neural Network
论文作者
论文摘要
冻结铸件已越来越多地用于处理各种多孔材料。其他研究人员报告了悬浮液中最终孔隙率与初始固体材料分数之间的线性关系。然而，孔隙率与固体材料之间的体积分数之间的关系在不同材料或冷冻溶剂之间显示出很高的差异，因为材料的性质显着影响冻结铸造中形成的孔。在这里，我们提出了一个人工神经网络（ANN），以评估冻结铸造过程中的孔隙率。经过良好的训练，ANN模型在实验数据上，可以从描述最具影响力过程条件的四个输入中预测孔隙率值。在本研究中还分析和讨论了模型的误差范围，可靠性和优化。结果证明，该方法有效地总结了一个模型中各种材料的一般规则，这很难通过线性拟合来实现。最后，还提供了基于训练有素的ANN模型的用户友好的迷你程序，以预测定制的冻结实验的孔隙率水平。
Freeze casting has been increasingly applied to process various porous materials. A linear relationship between the final porosity and the initial solid material fraction in the suspension was reported by other researchers. However, the relationship of the volume fraction between the porosity and the solid material shows high divergence among different materials or frozen solvents, as the nature of materials significantly affects the pores formed in freeze casting. Here, we proposed an artificial neural network (ANN) to evaluate the porosity in freeze casting process. After well training the ANN model on experimental data, a porosity value can be predicted from four inputs which describe the most influential process conditions. The error range, reliability and optimization of the model were also analyzed and discussed in this study. The results proved that this method effectively summarizes a general rule for diverse materials in one model, which is difficult to be realized by linear fitting. Finally, a user-friendly mini program based on a well-trained ANN model is also provided to predict the porosity level for customized freeze-casting experiments.
