|
基于卷积神经网络的短时傅里叶变换
|
Abstract:
本研究提出了一种基于卷积神经网络(CNN)的短时傅里叶变换方法,用于揭示非稳态信号的频谱随时间的变化规律。我们设计了一个卷积神经网络,采用双层结构,并随机初始化网络权重系数,不包含偏置系数。输入数据为随机生成的一维信号,其傅里叶变换作为标签数据。通过使用平方误差作为损失函数,并运用梯度下降法对网络进行训练,网络逐渐学得输入信号到其傅里叶变换的映射规则。同时,我们观察到网络权重系数在迭代过程中逐渐逼近傅里叶变换的核函数。基于学到的核函数,我们进行了信号的时频分析。数值试验结果表明,以通过训练获得的核函数作为基函数的短时傅里叶变换能取得与传统窗口傅里叶变换相一致的结果,证明了该方法的有效性。这一基于卷积神经网络的短时傅里叶变换方法为处理非稳态信号提供了一种新颖而有效的途径。
This study proposes a windowed Fourier transform method based on the Convolutional Neural Network (CNN) to reveal the temporal evolution of the spectrum of non-stationary signals. We designed a CNN with a dual-layer structure, randomly initialized network weight coefficients, and no bias terms. The input data consists of randomly generated one-dimensional signals, with their Fourier transforms serving as label data. Using the mean square error as the loss function and applying gradient descent for network training, the network gradually learns the mapping rules from input signals to their Fourier transforms. Simultaneously, we observed that the network weight coefficients gradually approximate the Fourier transform kernel during the iterative process. Based on the learned kernel function, we conducted a time-frequency analysis of signals. Numerical experimental results demonstrate that the windowed Fourier transform obtained through the learned kernel function as basis functions achieve consistent results with the traditional windowed Fourier transform, confirming the effectiveness of the proposed method. This CNN-based windowed Fourier transform method provides a novel and effective approach for processing non-stationary signals.
[1] | 武国宁, 曹思远, 马宁, 刘建军. S变换的时频分析特性及其改进[J]. 地球物理学进展, 2011, 26(5): 1661-1667. |
[2] | 武国宁, 曹思远, 孙娜. 基于复数道地震记录的匹配追踪算法及其在储层预测中的应用[J]. 地球物理学报, 2012, 55(6): 2027-2034. |
[3] | 周剑雄, 陈付彬, 石志广, 付强. 补零离散傅里叶变换的插值算法[J]. 信号处理, 2007(5): 690-694. |
[4] | 张德干, 郝先臣, 高光来, 赵海. 一种基于快速傅里叶变换的小波变换方法[J]. 东北大学学报, 2000(6): 598-601. |
[5] | 孟小红, 郭良辉, 张致付, 李淑玲, 周建军. 基于非均匀快速傅里叶变换的最小二乘反演地震数据重建[J]. 地球物理学报, 2008, 51(1): 235-241. |
[6] | Liu, Y. and Fomel, S. (2013) Seismic Data Analysis Using Local Time-Frequency Decomposition. Geophysical Prospecting, 61, 516-525. https://doi.org/10.1111/j.1365-2478.2012.01062.x |
[7] | Liu, G.C., Fomel, S. and Chen, X.H. (2011) Time-Frequency Analysis of Seismic Data Using Local Attributes. Geophysics, 76, 23-34. https://doi.org/10.1190/geo2010-0185.1 |
[8] | Wu, G.N., Fomel, S. and Chen, Y.K. (2018) Data-Driven Time-Frequency Analysis of Seismic Data Using Non-Stationary Prony Method. Geophysical Prospecting, 66, 85-97. https://doi.org/10.1111/1365-2478.12530 |
[9] | EI-Bakry, H.M. and Zhao, Q.F. (2004) Fast Object/Face Detection Using Neural Networks and Fast Fourier Transform. International Journal of Signal Processing, 1, 182-187. |
[10] | Li, Z.Y., Kovachki, N., Azizzadenesheli, K., et al. (2020) Fourier Neural Operator for Parametric Partial Differential Equa-tions. arXiv: 2010.08895. |
[11] | 杨庭威, 曹丹平, 杜南樵, 崔荣昂, 南方舟, 徐亚, 梁策. 基于深度学习的接收函数横波速度预测[J]. 地球物理学报, 2022, 65(1): 214-226. |
[12] | 张逸伦, 喻志超, 胡天跃, 何川. 基于U-Net的井中多道联合微地震震相识别和初至拾取方法[J]. 地球物理学报, 2021, 64(6): 2073-2085. |
[13] | 宋欢, 毛伟建, 唐欢欢. 基于深层神经网络压制多次波[J]. 地球物理学报, 2021, 64(8): 2795-2808. |
[14] | 赵明, 陈石, Yuen, D. 基于深度学习卷积神经网络的地震波形自动分类与识别[J]. 地球物理学报, 2019, 62(1): 374-382. |
[15] | Mathieu, M., He-naff, M. and LeCun, Y. (2013) Fast Training of Convolutional Networks through FFTs. arXiv: 1312.5851. |
[16] | Highlander, T. and Rodriguez, A. (2016) Very Efficient Training of Convolutional Neural Networks Using Fast Fourier Transform and Overlap-and-Add. arXiv: 1601.06815. |
[17] | Lin, S., Liu, N., Nazemi, M., Li, H.J., Ding, C.W., Wang, Y.Z. and Pedram, M. (2018) FFT-Based Deep Learning Deployment in Embedded Systems. 2018 Design, Automation & Test in Europe Conference & Exhibition (DATE), Dresden, 19-23 March 2018, 1045-1050. https://doi.org/10.23919/DATE.2018.8342166 |
[18] | Chitsaz, K., Hajabdollahi, M., Karimi, N., Samavi, S. and Shi-rani, S. (2020) Acceleration of Convolutional Neural Network Using FFT-Based Split Convolutions. arXiv: 2003.12621. |
[19] | Koplon, R. and Sontag, E.D. (1997) Using Fourier-Neural Recurrent Networks to Fit Sequential In-put/Output Data. Neurocomputing, 15, 225-248. https://doi.org/10.1016/S0925-2312(97)00008-8 |
[20] | Zhang, J., Lin, Y.B., Song, Z. and Dhillon, I.S. (2018) Learning Long Term Dependencies via Fourier Recurrent Units. arXiv: 1803.06585. |
[21] | Zhang, Y. and Chan, L.W. (2000) ForeNet: Fourier Recurrent Networks for Time Series Prediction. 7th International Conference on Neural Information Processing, 576-582.
https://api.semanticscholar.org/CorpusID:17884060 |
[22] | Cheng, Y., Yu, F.X., Feris, R.S., Kumar, S., Choudhary, A. and Chang, S.F. (2015) An Exploration of Parameter Redundancy in Deep Neural Networks with Circulant Projec-tions. IEEE International Conference on Computer Vision, Santiago, 7-13 December 2015, 2857-2865. https://doi.org/10.1109/ICCV.2015.327 |
[23] | Choromanski, K., Likhosherstov, V., Dohan, D., Song, X.Y., Davis, J., Sarlos, T., Belanger, D., Colwell, L. and Weller, A. (2020) Masked Language Modeling for Proteins via Linearly Scalabel Long-Context Transforms. arXiv: 2006.03555. |
[24] | Lee-Thorp, J., Ainslie, J., Eckstein, I. and Ontanon, S. (2021) FNet: Mixing Tokens with Fourier Transforms. arXiv: 2015.03824. |