从盲声源信号的独立性出发!提出了一种新的盲声源混合信号分离方法:该方法基于信号联合概率的

分布统计!利用信号联合概率的方向导数熵最小获得最佳的旋转角度!最终实现盲信号分离:与快速独立分

量分析方法及神经网络方法相比!该方法不需要迭代计算:采用新的盲声源信号分离方法对轴承试验台的混

合声音信号进行识别!将电机和滚动轴承的声音分离出来!进而可以准确识别机械的故障-Blind sound source from the independence of the starting signal! Proposed a new mixed-signal blind sound source separation method: This method is based on joint probability distribution of signal statistics! Use of signal joint probability of directional derivative entropy minimum get the best rotation angle ! finally realize blind signal separation: with the fast independent component analysis and neural network methods! This method does not require iterative calculation: the introduction of a new blind signal separation method of sound source on the bearing test bed of mixed sound signals to identify! to motor and rolling bearings separated voice! which can accurately identify the mechanical failure