| 摘要: |
| 针对非均质涂层组织均匀性超声衰减法表征中存在的非线性和不适定问题，提出一种基于多尺度超声衰减系数的粒子群优化-支持向量回归(PSO-SVR)表征方法。基于非均质材料中超声波散射的“多尺度效应”，利用连续小波变换获得涂层的多尺度超声衰减系数，使涂层在不同频带范围内的超声响应得到充分提取。以多尺度衰减系数作为输入向量，借助 SVR 在小样本条件下优异的数据挖掘和自动学习能力，实现多因素耦合约束下涂层组织均匀性信息的有效解耦，并引入粒子群优化和交互检验技术对 SVR 关键超参数进行全局优选。采用该模型对铝硅聚苯酯封严涂层的组织均匀性进行预测，结果表明， 涂层分布均匀性长度模型预测值与显微 CT 原位标定值间的决定系数 R2和均方误差 MSE 分别为 0.834 和 0.824，与反向传播算法(BP)、径向基神经网络(RBF)和广义回归神经网络(GRNN)等人工神经网络模型相比，PSO-SVR 模型在小样本条件下具有更好的泛化能力和更高的预测精度。研究结果为非均质材料组织均匀性的定量无损表征提供了新的研究思路。 |
| 关键词: 非均质涂层 组织均匀性 多尺度超声衰减系数 粒子群优化 支持向量回归 |
| DOI：10.11933/j.issn.1007?9289.20220118001 |
| 分类号:TB559 |
| 基金项目:国家自然科学基金(52075078)、企业创新青年人才托举计划(2021-1-2)和西安交通大学机械结构强度与振动国家重点实验室开放课题(SV2021-KF-01)资助项目 |
|
| Ultrasonic Quantitative Characterization of Heterogeneous Coating Microstructure Uniformity Based on PSO-SVR |
|
ZHANG Wei1,2, LIN Li3, FAN Junling1,2, MA Zhiyuan3
|
|
1.Aircraft Strength Research Institute, Xi’ an 710065 , China;2.State Key Laboratory of Structural Analysis for Industrial Equipment,Dalian University of Technology, Dalian 116024 , China;3.School of Materials Science and Engineering, Dalian University of Technology, Dalian 116024 , China
|
| Abstract: |
| In view of the nonlinear and ill-posed problems in the ultrasonic inversion of microstructure uniformity parameters, this paper proposes a PSO-SVR prediction model based on multi-scale ultrasonic attenuation coefficient. To decouple the “multi-scale scattering effect” of ultrasonic propagation in abradable seal coating, the echo signals are decomposed using the continuous wavelet transform (CWT). The ultrasonic responses in different frequency bands could be sufficiently extracted through the multi-scale ultrasonic attenuation coefficient obtained by CWT. Subsequently, taking the coefficients as input vectors, the SVR model is established. The parameters of SVR are optimized through PSO algorithm. The microstructure uniformity of the AlSi-polyester seal coating are predicted using the new model. The results show that, the R2 and MSE between the predicted uniformity length of the model and the in situ calibration values of the micro CT were 0.834 and 0.824 respectively. The comparison results show that the PSO-SVR model has higher accuracy, better generalization ability, and stronger robustness compared with classical ANN models (BP,RBF and GRNN), in the case with limited experimental data. This paper provides a new idea for quantitative nondestructive characterization of microstructure uniformity of heterogeneous materials. |
| Key words: heterogeneous coating microstructure uniformity multi-scale ultrasonic attenuation coefficient particle swarm optimization support vector regression |
