Combination of Wavelet Packet Analysis with BPNN Flaw Type Identification in Concrete Ultrasonic Testing

According to the nonstationarity of ultrasonic signals in concrete ultrasonic testing, a method of flaw classification was presented based on the combination of wavelet packet analysis and artificial neural network (ANN). The wavelet packet analysis is used to extract characteristic values reflecting the flaw properties and back propagation neural network(BPNN) is used to classify the characteristic values. An experiment system was used to test the method, in which 5 flaws were processed. The test results show that by this method human effects on qualitative recognition of flaws can be reduced and high accuracy of flaw classification can be obtained.