Micro-Cracks Detection of Eggshells Based on a Magnetostrictive Transducer

With respect to the problem that it is difficult to pick out eggs with small cracks, a novel vibrating type of eggshell cracks detection system using a magnetostrictive transducer was developed firstly. Then the acoustic signals was recorded while the transducer vibrating with the swept frequency from 1 to 14000 Hz. Thirdly, the wavelet and Burg power spectrum were used for data pretreatment and feature extraction respectively. By comparing the spectrums of intact eggs and hairline cracked eggs, the characteristic frequency of 4500~8000 Hz was found. Moreover, five feature parameters were defined for further analysis. In addition, the Support Vector Machine (SVM) based classifier was applied to build the model for identifying the cracked eggs from intact eggs. The experimental results show that the useful information of the acoustic signals was enhanced significantly and the identification accuracy reaches to 100 % and 98 % for training set and testing set respectively. This paper proposes a novel method for discriminating eggshell cracks using a magnetostrictive transducer in conjunction with the signal power spectrum analysis and LIBSVM methods which results show its feasibility. Copyright © 2015 IFSA Publishing, S. L.