Data mining technique of Acoustic Emission signals under supervised and unsupervised mode

Acoustic Emission (AE) can be used to discriminate the different types of damage occurring in a constrained metal material. However, the main problem associated with data analysis is the discrimination between the different acoustic emission sources, especially in a high-noise/interference environment. In this paper, cluster analysis, an important tool for investigating and interpreting data, was used to extract crack related signals from noise. More over, different kinds of noise signals were also classified successfully. On the basis of clustering analysis, the training samples quality of BP neural network was improved, also was the result of training. Well trained BP neural network has potential for a continuous on-line monitoring procedure to distinguish the initiation of severe damage from the AE signal even in a high-noise/ interference environment.