基于多输出支持向量回归的SHS-熔铸涂层性能模拟

Simulation on the Performance of SHS-Fusion Casting Coating Based on the Multi-Output Support Vector Regression

鉴于离心-自蔓延高温合成制备陶瓷涂层配方中添加剂对陶瓷涂层性能影响的复杂性,采用遗传算法优化支持向量机参数,建立添加剂与陶瓷涂层性能之间的支持向量回归模型,并用测试数据验证模型。结果表明,复合钢管陶瓷压溃强度及陶瓷涂层致密度的模拟值与试验值的相对误差最大值分别为6.2%和5.2%,采用遗传算法优化多输出支持向量机参数可以有效地提高模型的精度,为陶瓷涂层配方的优化提供新途径。

Considering the complexity between addictives in the formula and the performance of the ceramic coating prepared by centrifugal self-propagating high-temperature synthesis process（SHS）, the genetic algorithm was used to optimize the parameters of support vector machine, and a model which described the relationship between addictives and the ceramic performance was established by support vector regression. Besides, the model was verified by the tested data. The results show that the maximum relative error of the crushing strengthen of the compound steel-ceramic pipes and the relative density of the ceramic reach 6.2 % and 5.2 %, respectively, and the accuracy of the model was improved effectively with the parameters of M-SVR model optimized by the genetic algorithm, providing a optimization ways to the formula of ceramic coating.