摘要
研究水声基阵波束形成器的鲁棒性问题。由于水声信号的复杂性以及波束形成方法在处理水声基阵的实际问题时存在各种误差,为提高波束形成器的鲁棒性,提出一种改进SVM的方法。首先在支持向量回归运算模型中引入加载量,再利用拉格朗日法来对含有Hubert损失函数的回归模型进行计算,并将改进方法应用于水声基阵,不仅可以减小运算量,同时可以提高波束形成器的鲁棒性。最后通过计算机仿真和水池实验数据验证,并与加载采样矩阵求逆(Loading Sample MatrixInversion,LSMI)方法进行对比。实验结果表明:在无失配、水声基阵接收信号的方向存在偏差以及基阵阵元存在轻微扰动的各种情况下,都具有比LSMI强的鲁棒性。
Robustness of underwater array beamformer was studied in this paper. Because complexity of underwa- ter acoustic signal and mismatch with processing underwater acoustic array, how to improve the robustness of the beamformer is particularly important. This paper applied a Support Vector Machine (SVM) to the array beamforming to solve the problem. The loading force was introduced in the computation of support vector regression. Regression model contained Hubert loss function was calculated by Lagrangian method. Then the method was applied to under- water acoustic arrays. Finally, the method was validated by simulations and pool experimental data. Experimental re- sults show that the SVM - based beamforming method enhances the robustness better than Loading Sample Matrix In- version (LSMI) in terms of desired signal array manifold vector errors in an ideal scenario of no - mismatch and an actual scenario of mismatch respectively.
出处
《计算机仿真》
CSCD
北大核心
2013年第6期335-339,共5页
Computer Simulation
关键词
水声基阵
鲁棒性
支持向量机
加载采样矩阵求逆
Underwater acoustic array
Robust
Support vector machine ( SVM )
Loading sample matrix inversion (LSMI)