利用完全复数极限学习机增强近场声全息空间分辨率
|
摘要
针对稀疏测量阵列条件下近场声全息重建结果空间分辨率不足的问题,提出了一种基于完全复数极限学习机的全息声压插值方法。该方法首先将已测量的全息面复声压和对应的测点坐标组成训练样本输入完全复数极限学习机,接着把插值点的坐标代入训练好的极限学习机,得到相应位置的复声压,实现全息数据的插值。利用插值后的全息数据进行重建,并与不做插值处理的重建结果和传统插值处理后的重建结果比较。仿真和实验结果均表明:与不做插值相比,该方法在不增加传声器的条件下显著提高了重建结果的空间分辨率。与基于支持向量机或传统极限学习机的插值方法相比,该方法速度更快,插值后重建结果精度更高。同时,通过添加噪声干扰验证了该方法的稳健性。
For the problem of insufficient spatial resolution of nearfield acoustic holography reconstruction under the sparse measurement array condition, a hologram pressure interpolation method based on the fully complex extreme learning machine is proposed. Firstly, the measured complex hologram pressures and the planar coordinates of the corresponding measurement points are combined into the training samples. They are used to train the fully complex extreme learning machine. Then, the planar coordinates of the interpolation points are introduced into the trained machine to generate the corresponding pressures, achieving the interpolation of hologram data. Finally, the interpolated hologram pressures are used to reconstruct. The results are compared with those without the interpolation and with the traditional interpolation. Numerical simulations and experimental results show that compared to the results without the interpolation, the proposed method significantly improves the spatial resolution of the reconstructed results without increasing the microphones. Compared to the interpolation method based on the support vector machine or the traditional extreme learning machine, the proposed method is faster and its reconstructed result is more accurate. The robustness of this method is also verified by adding noise interference.
For the problem of insufficient spatial resolution of nearfield acoustic holography reconstruction under the sparse measurement array condition, a hologram pressure interpolation method based on the fully complex extreme learning machine is proposed. Firstly, the measured complex hologram pressures and the planar coordinates of the corresponding measurement points are combined into the training samples. They are used to train the fully complex extreme learning machine. Then, the planar coordinates of the interpolation points are introduced into the trained machine to generate the corresponding pressures, achieving the interpolation of hologram data. Finally, the interpolated hologram pressures are used to reconstruct. The results are compared with those without the interpolation and with the traditional interpolation. Numerical simulations and experimental results show that compared to the results without the interpolation, the proposed method significantly improves the spatial resolution of the reconstructed results without increasing the microphones. Compared to the interpolation method based on the support vector machine or the traditional extreme learning machine, the proposed method is faster and its reconstructed result is more accurate. The robustness of this method is also verified by adding noise interference.
引文
1 Maynard J D, Williams E G, Lee Y. Nearfield acoustic holography:I. Theory of generalized holography and the development of NAH. J. Acoust. Soc. Am., 1985; 78(4):1395-1413
2 Veronesi W A, Maynard J D. Near-field acoustic holography(NAH):II. Holographic reconstruction algorithms and computer implementation. J. Acoust. Soc. Am., 1987;81(5):1307-1322
3 Scholte R. Fourier based high-resolution near-field sound imaging. PhD thesis, Eindhoven University of Technology,2008
4 Harris M C, Blotter J D. Obtaining the complex pressure field at the hologram surface for use in near-field acoustical holography when pressure and in-plane velocities are measured. J. Acoust. Soc. Am., 2006; 119:809-816
5徐亮,毕传兴,陈心昭等.基于带限信号恢复算法的近场声全息分辨率增强方法.科学通报,2008; 53(14):1632-1639
6徐亮,陈心昭,毕传兴等.近场声全息分辨率增强的正交球面波插值方法.浙江大学学报(工学版),2009; 43(10):1808-1811
7张小正,毕传兴,徐亮等.基于波叠加法的近场声全息空间分辨率增强方法.物理学报,2010; 59(8):5564-5570
8 Kotsiantis S B, Zaharakis I D, Pintelas P E. Machine learning:a review of classification and combining techniques.Artificial Intelligence Review, 2006; 26(3):159-190
9 Jordan M I, Mitchell T M. Machine learning:trends, perspectives, and prospects. Science, 2015; 349(6245):255
10陈东成.基于机器学习的目标跟踪技术研究.博士论文,中国科学院大学,2015
11周志华.机器学习.北京:清华大学出版社,2016
12毛荣富,朱海潮,张劲松.近场声全息(NAH)中减少测量点数的研究.声学技术,2009; 28(3):287-294
13毛荣富,朱海潮,柳志忠等.利用支持向量回归插值减少全息面测量点数方法的试验研究.振动与冲击,2010; 29(3):128-131
14杜向华,朱海潮,毛荣富,朱海鹏.基于支持向量回归的patch近场声全息研究.声学学报,2012; 37(3):286-293
15 Bhavsar H, Panchal M H. A review on support vector machine for data classification. International Journal of Advanced Research in Computer Engineering&Technology,2012; 1(10):185-189
16 Huang G B,Zhu Q Y, Siew C K. Extreme learning machine:a new learning scheme of feedforward neural networks.IEEE International Joint Conference on Neural Networks,2004; 2:985-990
17 Huang G B, Zhu Q Y, Siew C K. Extreme learning machine:theory and applications. Neurocomputing, 2006; 70:489-501
18孙超,何元安,商德江等.全息数据外推与插值技术的极限学习机方法.哈尔滨工程大学学报,2014; 35(5):544-551
19 Kim T. Fully complex multilayer perception and its application to communications. PhD thesis, University of Maryland, 2002
20 Kim T, Adali T. Approximation by fully complex multilayer perceptrons. Neural Comput.,2003; 15(7):1641-1666
21 Li M B, Huang G B, Saratchandran P et al. Fully complex extreme learning machine. Neurocomputing, 2005; 68:306-314
22 Haykin S.神经网络与机器学习.北京:机械工业出版社,2016
23 Zhou H F, Lopez-Arteaga I, Nijmeijer H. Broadband planar nearfield acoustic holography based on one-third-octave band analysis. Applied Acoustics, 2016; 109:18-26
2 Veronesi W A, Maynard J D. Near-field acoustic holography(NAH):II. Holographic reconstruction algorithms and computer implementation. J. Acoust. Soc. Am., 1987;81(5):1307-1322
3 Scholte R. Fourier based high-resolution near-field sound imaging. PhD thesis, Eindhoven University of Technology,2008
4 Harris M C, Blotter J D. Obtaining the complex pressure field at the hologram surface for use in near-field acoustical holography when pressure and in-plane velocities are measured. J. Acoust. Soc. Am., 2006; 119:809-816
5徐亮,毕传兴,陈心昭等.基于带限信号恢复算法的近场声全息分辨率增强方法.科学通报,2008; 53(14):1632-1639
6徐亮,陈心昭,毕传兴等.近场声全息分辨率增强的正交球面波插值方法.浙江大学学报(工学版),2009; 43(10):1808-1811
7张小正,毕传兴,徐亮等.基于波叠加法的近场声全息空间分辨率增强方法.物理学报,2010; 59(8):5564-5570
8 Kotsiantis S B, Zaharakis I D, Pintelas P E. Machine learning:a review of classification and combining techniques.Artificial Intelligence Review, 2006; 26(3):159-190
9 Jordan M I, Mitchell T M. Machine learning:trends, perspectives, and prospects. Science, 2015; 349(6245):255
10陈东成.基于机器学习的目标跟踪技术研究.博士论文,中国科学院大学,2015
11周志华.机器学习.北京:清华大学出版社,2016
12毛荣富,朱海潮,张劲松.近场声全息(NAH)中减少测量点数的研究.声学技术,2009; 28(3):287-294
13毛荣富,朱海潮,柳志忠等.利用支持向量回归插值减少全息面测量点数方法的试验研究.振动与冲击,2010; 29(3):128-131
14杜向华,朱海潮,毛荣富,朱海鹏.基于支持向量回归的patch近场声全息研究.声学学报,2012; 37(3):286-293
15 Bhavsar H, Panchal M H. A review on support vector machine for data classification. International Journal of Advanced Research in Computer Engineering&Technology,2012; 1(10):185-189
16 Huang G B,Zhu Q Y, Siew C K. Extreme learning machine:a new learning scheme of feedforward neural networks.IEEE International Joint Conference on Neural Networks,2004; 2:985-990
17 Huang G B, Zhu Q Y, Siew C K. Extreme learning machine:theory and applications. Neurocomputing, 2006; 70:489-501
18孙超,何元安,商德江等.全息数据外推与插值技术的极限学习机方法.哈尔滨工程大学学报,2014; 35(5):544-551
19 Kim T. Fully complex multilayer perception and its application to communications. PhD thesis, University of Maryland, 2002
20 Kim T, Adali T. Approximation by fully complex multilayer perceptrons. Neural Comput.,2003; 15(7):1641-1666
21 Li M B, Huang G B, Saratchandran P et al. Fully complex extreme learning machine. Neurocomputing, 2005; 68:306-314
22 Haykin S.神经网络与机器学习.北京:机械工业出版社,2016
23 Zhou H F, Lopez-Arteaga I, Nijmeijer H. Broadband planar nearfield acoustic holography based on one-third-octave band analysis. Applied Acoustics, 2016; 109:18-26