Classifying ships by their acoustic signals with a cross-bispectrum algorithm and a radial basis function neural network

Abstract

An algorithm for estimating the cross-bispectrum of an acoustic vector signal was formulated. Composed features of sound pressure and acoustic vector signals are extracted by the proposed algorithm and other estimating algorithms for secondary and higher order spectra. Its effectiveness was tested with lake and sea trial data. These features can be used to construct an input vector set for a radial basis function neural network. The classification of vessels can then be made based on the extracted features. It was shown that the composed features of acoustic vector signals are more easily divided into categories than those of pressure signals. When using the composed features of acoustic vector signals, the recognition rate of underwater acoustic targets improves.

Keywords

acoustic vector signal / cross-bispectrum / feature extraction / RBFNN / ship classification

References

[1] Zhang Xianda. Linear algebra in signal processing[M]. 1997, Beijing: Science Press

[2] Sun Guiqing. The study of detection technology based on the vector hydrophone[D]. 2001, Harbin: Harbin Engineering University

[3] Li Chunxu. Combined signal processing technology with acoustic pressure and particle velocity[D]. 2000, Harbin: Harbin Engineering University

[4] Li S., Yang Desen. Comparing sound pressure signals and acoustic vector signals by nonintegral dimension spectrum[J]. Journal of Harbin Engineering University, 2007, 28(5): 510-514

[5] Mendel J. M. Tutorial on higher-order statistics (spectra) in signal processing and system theory: theoretical results and some applications[J]. Proceedings of the IEEE, 1991, 79(3): 278-305

[6] Nikias C. L., Mendel J. M. Signal processing with higher-order spectra[J]. IEEE Signal Processing Magazine, 1993, 10(3): 10-37

[7] WILAMOWSKI B M, JAEGER R C. Implementation of RBF type networks by MLP networks[C]// Proceedings of the IEEE International Conference on Neural Networks. Piscataway, 1996: 1670–1675.

[8] Fang S., Lu Jiren. The application of a combined basis neural network in the underwater acoustic target classification and recognition[J]. Technical Acoustics, 1998, 17(2): 54-62

[9] Peng J., Li K., Huang Deshuang. A hybrid forward algorithm for RBF neural network construction[J]. IEEE Trans on Neural Networks, 2006, 17(6): 1439-1451

[10] Panchapakesan C., Palaniswami M., Ralph D., et al. Effects of moving the centers in an RBF network[J]. IEEE Trans on Neural Networks, 2002, 13(6): 1299-1307

[11] Peng H., Ozaki T., Haggan-Ozaki V., et al. A parameter optimization method for radial basis function type models[J]. IEEE Trans on Neural Networks, 2003, 14(2): 432-438

[12] Er M. J., Chen W., Wu S. High-speed face recognition based on discrete cosine transform and RBF neural networks[J]. IEEE Trans on Neural Networks, 2005, 16(3): 679-691

[13] Gonzalez J., Rojas I., Ortega J., et al. Multiobjective evolutionary optimization of the size, shape, and position parameters of radial basis function networks for function approximation[J]. IEEE Trans on Neural Networks, 2003, 14(6): 1478-1495

[14] DING S Q, XIANG C. From multilayer perceptrons to radial basis function networks: a comparative study[C]// Proceedings of the 2004 IEEE Conference on Cybernetics and Intelligent Systems. New York, 2004: 69–74.