FCM-FastICA的点蚀声发射信号分离识别方法*

金属点蚀是一种破坏性和隐患较大的设备损伤形式。点蚀会产生声发射信号。点蚀过程中产生的多种声源类型会造成信号混叠，影响腐蚀进程的判断。针对点蚀信号混叠问题，提出一种模糊C均值聚类与快速独立分量分析算法相结合的点蚀信号分离识别方法。通过分析单、双点蚀声发射数据将点蚀分为钝化膜破裂阶段、点蚀诱导成核及发展阶段，由聚类确定信号类别并用快速独立分量分析分离混合信号，利用相关性函数验证分离效果。结果表明：单点蚀过程存在3类原信号，双点蚀过程存在7类信号，其中包含单个信号与混合信号；单个信号与原信号相关性极高，达到0.8以上，混合信号的分离分量与原信号相关性达到0.6以上，分离效果较好。该方法可对点蚀混合信号进行有效分离和识别，为腐蚀进程判断提供支持。

Pitting acoustic emission signal separation and recognition method based on FCM-FastICA

Metal pitting is a destructive and hidden damage form of equipment, which will generate acoustic emission signals. The various sound source types generated in the pitting process will cause signal aliasing and affect the judgment of corrosion process. Aiming at the problem of pitting signal aliasing, a pitting signal separation and recognition method based on FCM clustering and FastICA algorithm was proposed. By analyzing single and double pitting AE data, the pitting process was divided into passive film rupture stage, pitting induced nucleation stage and development stage. The signal categories were determined by clustering and the mixed signals were separated by FastICA, and the separation effect was verified by correlation function. The results show that there are three kinds of original signals in single pitting process and seven kinds of signals in double pitting process, including single signal and mixed signal. The correlation between a single signal and the original signal is extremely high, reaching above 0.8. The correlation between the separation component of the mixed signal and the original signal is more than 0.6. The separation effect is good. This method can effectively separate and identify the pitting mixture signals and provide support for judging the corrosion process.