基于支持向量机的夹层结构水声性能预测建模  

Modeling the Underwater Acoustic Properties of Sandwich Plates Based on Support Vector Machines

作　　者：李永清[1] 朱锡[1] 石勇[1] 李爽[2] 

机构地区：[1]海军工程大学船舶与海洋工程系,武汉430033 [2]海军装备部综合计划部,北京100038

出　　处：《武汉理工大学学报》2007年第7期99-102,共4页Journal of Wuhan University of Technology

摘　　要：支持向量机是近几年发展起来的机器学习的新方法,它较好地解决了小样本、非线性、高维数、局部极小点等实际问题。选用典型的三明治结构并制备玻璃钢/吸声橡胶/玻璃钢夹层复合材料,通过改变吸声橡胶层的厚度,得到一系列夹层复合吸声材料试件,并在水声材料声脉冲管测试系统中测得试件的吸声系数和声压反射系数。在这些实验数据的基础上,利用最小二乘支持向量机(LS-SVM)研究了材料不同厚度和频率与其吸声和反射系数对应关系的建模,结果表明最小二乘支持向量机用于材料水声性能的预测建模是一种非常有效的方法。Support vector machine （SVM） was a novel machine learning method, which was powerful for the problem characterized by small sample, nonlinearity, high dimension and local minima, and has high generalization. In this paper, the least square support vector machine（LS-SVM）, which developed from normal SVM, has been used for predicting and modeling the underwater acoustic properties of sandwich plates, which were composed of fiberglass-reinforced plastics/acoustic rubber/fiberglass-reinforced plastics. Effective result indicated that LS-SVM was of potential application in acoustic material research.

关 键 词：最小二乘支持向量机 三明治结构 水声性能 

分 类 号：TQ050.43[化学工程]