改进盲源信号处理的集成动力装置故障诊断

An Improved Blind Source Separation Method for the Fault Diagnosis of Integrated Power Unit

针对高速集成动力装置中振动信号相互耦合严重而普通方法无法进行有效分离的问题,提出一种改进的盲源信号处理方法.该方法在传统盲源分离方法的基础上,通过故障机理获得的先验知识确定源的数量和频率特征,从而有效地提升故障信号源的分离效果.试验验证结果表明,采用改进的盲源信号处理方法处理后的信号与实际系统的故障特征吻合度很高,该方法能够有效地应用于高速集成动力装置的故障诊断并提高诊断精度.

In high-speed integrated power units, vibration signal coupling is very serious, which makes it difficult to determine the number of vibration sources by using traditional methods. In order to cope with this problem, this paper proposed an improved blind source signal processing method. Based on the traditional blind source separation method, this method used prior knowledge of failure mechanism to enhance the effect of blind source separation. Experiment results show that signal after separation fits well with fault characteristics. This method is very effective in the fault diagnosis of integrated power unit and it can improve the accuracy of fault diagnosis.