陶瓷涂层材料多模态数据表征学习

陶瓷涂层具有耐高温、耐腐蚀、耐磨损等特性, 其热膨胀系数和热导率等参数与其性能息息相关. 为解决陶瓷涂层性能实验成本高、测试困难等问题, 提出了陶瓷涂层材料多模态数据表征学习的性能预测方法. 首先利用高斯混合模型虚拟样本生成(Gaussian mixture model virtual sample generation, GMMVSG)算法生成符合真实陶瓷涂层数据分布的样本来扩充数据集; 其次利用卷积神经网络 VGG16 对陶瓷涂层的显微结构图像数据进行特征提取, 利用 TabNet 对结构化数据进行特征提取, 将提取到的图像数据特征与结构化数据特征融合; 最终根据多模态数据表征建立基于K-最近邻(K-nearest neighbor, KNN)、支持向量机回归(support vector regression, SVR)和多层感知机(multi-layer perceptron, MLP) 3 种机器学习算法的预测模型, 对陶瓷涂层的性能指标, 即热膨胀系数和热导率进行了预测. 实验结果表明: 提出的多模态数据表征学习模型的预测结果要优于单模态数据表征学习模型, 其中基于 MLP 算法训练的多模态数据表征学习模型对陶瓷涂层性能的预测效果最好; 在测试集中, 对陶瓷涂层热膨胀系数预测的平均绝对误差(mean absolute error, MAE)和均方误差(mean square error, MSE)分别为 0.026 6 和 0.001 7, 对热导率预测的 MAE 和 MSE 分别为 0.017 9 和 0.000 7. 所提出的陶瓷涂层材料多模态数据表征学习方法有效融合了结构化数据与非结构化数据, 联合学习了各模态数据的潜在共享信息, 成功提升了对陶瓷涂料层材料性能预测的准确度.

Explosive synchronisation of multi-layer networks based on frequency-weighted coupled model

Synchronisation is a prevalent and universal dynamic behaviour of complex systems. Based on the critical coupling strength of explosive synchronisation in multi-layer complex networks, a frequency-weighted Kuramoto model was developed in this study. We obtained the critical coupling strength for explosive synchronisation through field tests, self-consistent analysis, and numerical simulation. The theoretical analysis and simulation results show that the critical value is related to the strength of the interlayer interaction and the average degree of the network. Increasing the interaction between layers prevents explosive synchronisation, whereas increasing the average degree of the network facilitates explosive synchronisation. The results are useful for studies on complex networks and systems.