Abstract: This paper applies Support Vector Machine (SVM) based on the rule of structural risk minimization to the reverberation time series prediction, and compares the predict results with Radial Basic Function(RBF) neural network. Experimental results of an ocean reverberation show that the SVM method is more effective than RBF and suitable to predict the reverberation time series.

Key words: shallow-water reverberation, Support Vector Machine(SVM), Radial Basic Function(RBF) neural network, prediction

摘要： 把基于结构风险最小化原则的支持向量机应用到混响时间序列预测中，与径向基函数(RBF)神经网络方法预测结果进行了对比分析。采用海上实验混响数据进行预测，处理结果表明，支持向量机的方法优于RBF神经网络的方法，对混响时间序列有很好的预测效果。

关键词: 浅海混响, 支持向量机, 径向基函数神经网络, 预测

CLC Number: 

• TP18