论文部分内容阅读
从盲源信号分离后非高斯性最大化出发,提出了一种基于经验特征函数的盲源信号分离方法.该方法把经验特征函数与概率密度函数一一对应,并以混合信号与高斯信号的经验特征函数的欧氏距离最大化作为判据,以固定点算法为优化算法进行盲源分离.该方法克服了FastICA算法中选取不同的近似函数对不同概率密度分布的信号效果不佳的问题.仿真实验结果表明,与常用的几种FastICA算法相比,该方法具有更好的收敛效果.采用新的盲源信号分离方法对管道破坏产生的实际声发射信号进行分离,可将破坏点互相关定位精度提高到3%以上.
Starting from the maximization of non-Gaussianity after blind source separation, a blind source separation method based on empirical eigenfunction is proposed. The method combines the empirical eigenfunction with the probability density function, and combines the mixed signal with the Gaussian signal , The Euclidean distance maximization of the empirical eigenfunction is used as the criterion and the fixed-point algorithm is used as the optimization algorithm for blind source separation. This method overcomes the problem that the different approximation functions in the FastICA algorithm have poor effect on signals with different probability density distributions The simulation results show that this method has better convergence compared with several popular FastICA algorithms.A new method of blind source signal separation is used to separate the actual AE signals from the pipeline damage, Related positioning accuracy increased to 3% or more.