Reconfigurable architecture of ultrasonic defect detection based on wavelet packet and back propagation artificial neural network

摘要

The defects detection based on algorithms of discrete wavelet packet transform and back propagation artificial neural network is described firstly. Then the reconfigurable architecture of the defects detection in embedded reconfigurable system is discussed. Upon the reconfigurable architecture, the algorithms of discrete wavelet packet transform and back propagation neural net work are rearranged. The reconfigurable architectures of both algorithms are carried out eventually. According to the experiments of ultrasonic signal processing, the reconfigurable architecture of the defect detection can provides a flexible and efficient solution to embedded reconfigurable signal processing system.