摘要
近场声全息(NAH)是一种进行噪声源分析的有效工具。但要获得较高的重建精度必须在全息面上大孔径范围内进行密集的数据采集,需要大量的测量点,测量任务十分繁重。因此在保证重建精度的前提下研究如何有效地减少全息面测量点数对NAH的意义十分重大。在总结前人工作的基础上,提出了一种利用支持向量回归对全息面进行插值的方法。该方法可以在全息面传声器数目较少时通过插值提高重建精度,因而可大幅地减少全息面的测量点数。对于两个矩形板声辐射数值计算中减少的测量点数都可高于70%,并得到了其它有意义的结论,达到了保证重建精度的前提下有效地减少全息面测量点数的研究目的。
Near-field Acoustic Holography （NAH） is a powerful tool in noise source analysis. However, accurate reconstruction requires dense measurement grid on large hologram, a huge number of microphones are demanded. It is significant to find a way to reduce the microphone number effectively while keeping high reconstruction accuracy in NAH implementation. On the basis of previous work, a hologram interpolation approach based on Support Vector Regression （SVR） is presented in this paper. Reconstruction accuracy can be improved significantly under small microphone number circumstance, therefore, a great amount of measurement points can be reduced by using the presented approach. In the numerical calculation of rectangle plane radiation, more than 70% measurement point reduction can be achieved in both two radiation cases, and other useful conclusions are also drawn.
出处
《声学技术》
CSCD
2009年第3期287-294,共8页
Technical Acoustics
关键词
近场声全息
支持向量回归
插值
重建精度
near-field acoustic holography, support vector regression, interpolation, reconstruction accuracy
