盲源分离在机械振动信号分析中的应用

Study on Blind Source Separation of Mechanical Vibration Signal

设备状态信号的处理是状态监测及故障诊断的基础。在实际运行环境中,信号检测传感器采集的机械振动信号必然包含设备各个部件的信号以及周围环境的强烈干扰。传统的振动信号处理方法抗扰去噪效果并不理想。盲源分离技术由于自身独特的盲处理优势,可以有效去除外来干扰并分离出源信号,有助于提高诊断的准确性。针对直升机齿轮箱振动信号进行盲源分离仿真,分离出了轴承故障振动信号,并将分离信号的功率谱与原始信号的功率谱相比较,表明盲源分离技术是机械故障诊断领域的一个有效的信号处理方法。

Equipment condition signal processing is the foundation of condition monitoring and fault diagnosis. In the actual environment, the mechanical vibration signals which the sensors gathered inevitably contain interference from other parts and equipments. The effect of interference removal by the traditional vibration signal processing methods is not adequate. However, blind sources separation is a special tool for analyzing and processing signals blindly, it can remove noises in observation signals effectively and improve the accuracy of the diagnostic performence. The application of BSS to diagnosis of helicopter gearboxes is proposed. One simulation is done on an actual faulty bearing and the bearing vibration signal is separated by BSS methods. Power spectrum densities of separated signal and original signal are compared and the results show that the method of blind sources separation can be a promis ing tool for analyzing and processing signals in condition monitoring and fault diagnosis of machinery.