有限元分析及SVR参数优化在SCFRW电特性中的应用

Application on finite element analysis and SVR parameters optimization of the electrical characteristics of SCFRW

复合材料特性的模型化仿真研究是工艺优化设计的重要辅助途径。本文以试验制备的短切碳纤维木质导电功能复合材料(SCFRW)作为检测样板,通过对样本不同参数下的电特性测试分析,构建了基于单元特性分析理论的有限元电特性仿真模型,并利用ANSYS软件进行了模拟仿真。同时,采用实验测得的表电阻率数据作为学习样本,利用支持向量回归机(SVR)算法对有限元仿真模型参数进行回归拟合并进行仿真预测,寻求有限元建模的最优参数。与实验数据的对照分析验证了SVR优化模型的可靠性和准确性,为实现连续变化参数下的有限元分析提供有效依据。利用SVR优化参数的有限元分析不仅为基于模型化的SCFRW复合材料的工艺优化与控制、材料制备提供科学指导,而且可达到减少材料消耗与浪费的预期目的。

Modeling simulation research is one of the important process optimization design aided approaches for composite materials. Short Carbon Fiber Reinforced Woodiness（SCFRW） electrical conducting functional composites are the testing samples in this paper. With the sample testing of electrical properties under different parameters, the finite element simulating models were built based on unit characteristic theory and simulated in ANSYS software.At the same time, surface resistivity data measured by the experiment as learning samples, regression fitting and prediction of the finite element simulating model parameters were implemented by using support vector regression（SVR） machine algorithm, so as to gain optimal finite element modeling parameters. The reliability and accuracy of SVR optimization model were verified by contrasting and analyzing with the experimental data, which provided effective basis for realizing finite element analysis under continuous variation. The research does not only provide scientific guidance for process modeling optimizational, control and material preparation, but also achieve the desired purpose of reducing material consumption and waste.