论文部分内容阅读
结合改进的固定点算法,解决了噪声环境下的ICA问题。根据噪声分布特性,分两个阶段去除不同类型的随机噪声。在预处理阶段去除了加性高斯白噪声,预处理后的数据采用改进的固定点算法,盲分离出有效信号和非高斯随机噪声。提出了对固定点算法迭代过程中设定较精确的初始值问题的算法,该方法能较为准确地设置初始值,使算法能提取有效信号。通过仿真实验和对实际地震数据的处理,得到了满意的分离结果,较好地恢复了有效信号。此外,当实际地震数据加载了较强噪声,信噪比降低时,采用本文算法进行盲分离,同样取得了良好的效果,再次验证了本文算法具有良好的稳健性和适应性。将盲分离算法应用到实际地震数据处理方面的研究,有助于地震资料的解释,同时这种处理技术的研究也能够促进盲分离技术的发展。
Combined with the improved fixed-point algorithm, the ICA problem in noisy environment is solved. According to the noise distribution characteristics, different types of random noise are removed in two stages. Additive Gaussian white noise is removed in the preprocessing stage. The improved fixed point algorithm is used for the preprocessed data to blindly separate the valid signal and the non-Gaussian random noise. An algorithm is proposed to set the exact initial value in iterative process of fixed-point algorithm. This method can set the initial value more accurately and make the algorithm extract the effective signal. Through the simulation experiment and the processing of the actual seismic data, the satisfactory separation result is obtained, and the effective signal is recovered well. In addition, when the actual seismic data is loaded with strong noise and the signal-to-noise ratio is reduced, the blind separation using the proposed algorithm also achieves good results. The algorithm verifies again that the proposed algorithm has good robustness and adaptability. The application of blind separation algorithm to the actual seismic data processing helps to explain the seismic data, and the research of this processing technology can also promote the development of blind separation technology.