Details

Autor(en) / Beteiligte

• Sun, Hong Chun

Titel

Pattern Recognition of the Steel Rod Based on Artificial Neural Network and SVM

Ist Teil von

• Advanced materials research, 2011, Vol.301-303, p.329-333

Ort / Verlag

Trans Tech Publications Ltd

Erscheinungsjahr

2011

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• The steel rod is an important part for project fields , and it is large-scale to be used. it is apt to crack, corrosion and so on in the poor working conditions. In order to recognize correctly the type of defects, a method was presented to extract frequency band energy feature by using wavelet package decomposition. In the meantime, to extract the peak-peak value in the time-domain and make the mixed feature vector. With the way of pattern recognition, the best recognition way was got by comparing the BP artificial neural network(ANN), PNN(probability neural network) artificial neural network and "one-versus-one" support vector machine(SVM).The result showed that the recognition rate of SVM was more suitable for defects’ identification in steel rod.