Recognition of a Sucker Rod´s Defect with ANN and SVM

In order to improve the recognition rate of a sucker rod´s defect and reduce the rapture possibility of the rod, the mixed characters include of wavelet packet energy character and the peak value in the time-domain were used as the input of a recognition network, and artificial neural networks (ANN) and support vector machines (SVM) were used and compared as the recognition network to get the best recognition way. Tested results with lots of data acquired in laboratory showed that SVM was better than ANN at recognition of the sucker rod´s defect, and SVM based on the mixed characters can enhance recognition rate of the sucker rod´s defect.