机械噪声故障特征提取的盲分离法与小波提纯法

Acoustic Feature Extraction of Rotating Machines Using Blind Source Separation and Wavelet Analysis

机械噪声故障特征提取的难点在于观测信号的信噪比较小 .将盲分离技术引入噪声故障特征提取 ,通过声源信号的相互独立性质 ,使用二阶盲分离算法从观测的混合信号中提取独立声源信号 ,然后 ,通过随机噪声与有效信号在多尺度空间中模极大值的不同传播特性 ,使用小波模极大值法提取有效信号特征 .该算法不仅消除了临近机器或部件辐射噪声的干扰 ,还消除了随机噪声的干扰 ,有效提取了机械噪声故障特征 .

In acoustic monitoring, the observed signal is usually the mixture of sound signals of all machines and it has a very low signal-to-noise ratio. To eliminate the mutual interference of sound signals, the blind source separation was used to recover the sound signals of independent sources. The second-order blind separation algorithm was proposed to reconstruct the spectrum of the monitored system. Then, the wavelet analysis was introduced and the maximum modulus method was used to remove the interference of random noise. The signal-to-noise ratio of monitored system is significantly enhanced via the proposed methods. The acoustic features can be obtained from the purified signal easily. The experiment results made in semi-anechoic chamber demonstrate the effectiveness of the presented methods.