盲源分离技术是信号处理领域的一个活跃的研究方向。人们发现可以将盲源分离问题纳入到贝叶斯网络的框架,利用变分贝叶斯理论通过对网络隐变量的学习得到源信号的估计。结合独立分量分析方法的研究历史,介绍了基于贝叶斯网络的盲源分离方法的相关进展。从盲解卷、非线性混合和噪声背景下的盲分离等方面对国内外研究进行了综述。 Blind source separation technology is an active research field in signal processing. People find that the problem of blind source separation can be included in the framework of Bayesian network, and the variation of Bayesian theory can be used to estimate the source signal by learning the hidden variables of the network. Combining with the history of research on independent component analysis, the progress of blind source separation based on Bayesian network is introduced. The researches at home and abroad are reviewed from the aspects of blind unwrapping, nonlinear mixing and blind separation under noisy background.