Research On DSP Technology In Pattern Recognition Of Acoustic Emission Signals

It will generate acoustic emission (AE) signals while material is distorted or broken under stress action. Through collecting and analyzing AE signals, we can detect and diagnose flaw or damage from the material. Our research group has been testing research on wood-plastic composite (WPC) AE signals for many years and we have achieved good results. In order to analyze and recognize the defect source's AE characteristics on-line, we design an online detection system based on DSP technology in this paper.Firstly, AE technique, wavelet transform theory and neural network pattern recognition theory are discussed. Further more, pattern recognition methods on AE signals of WPC is expounded based on the existing research results.Secondly, using wavelet packet transform theory, back propagation (BP) neural network and other algorithm, an AE signal pattern recognition model is designed based on Simulink. This model is applied to decompose AE signals using wavelet packet, to extract the characteristic parameters and to confirm the type corresponding to the flaw or damage at last.Thirdly, an AE signal pattern recognition hardware system is built using SEED-AD9243M high-speed data collection board and SEED-DEC6416 digital signal processing board. It provides hardware support for carrying out pattern recognition of AE signals.Finally, using high-speed code generation technology, an AE signal pattern recognition software system is designed. This software system is integrated by AE signals collection and analysis.The AE signal pattern recognition system designed in this paper can be used in WPC's flaw or damage on-line recognition and it is valid in practical application.