Damage Pattern Recognition of Refractory Materials Based on BP Neural Network

• Published in: Neural Information Processing pp. 431 - 440
• Main Authors: Liu, Changming, Wang, Zhigang, Li, Yourong, Li, Xi, Song, Gangbing, Kong, Jianyi
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2012
• Series: Lecture Notes in Computer Science
• Subjects: Acoustic emission; BP neural network; Pattern recognition; Refractory materials
• ISBN: 9783642344770; 3642344771
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-642-34478-7_53

The determination of the damage mode and the quantitative description of the damage of the clustered acoustic emission (AE) signal of the refractory materials based on the BP (back propagation) Neural Network are the subjects of this paper. In this paper, a large number of AE signals in the process of a three-point bending test were studied and the pattern recognition system of refractory materials based on BP neural network was established with the AE characteristic parameters such as amplitude, counts, rise time, duration and centroid frequency etc. The results show that the total recognition rate of material damage types with this method is as high as 97.5%, and the prediction error of the extent of the damage is about 5%, which indicates that this method has the value of application and dissemination in the aspect of micro-damage pattern recognition and extent prediction of the damage.