Study on Cracks Detection of Eggshells by Using Resonant Inspection Method and Identified by Support Vector Machine

• Main Authors: Pei-Hsuan Feng, 馮珮宣
• Other Authors: Ching-Wei Cheng
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/61800134023469273866

碩士 === 國立中興大學 === 生物產業機電工程學系所 === 105 === Cracks of eggshells will not only affect the preserved time, but also reduce the successful rate of the processed products. Therefore, this study will base on the theory of resonant inspection, and it was verified by the Support Vector Machine（SVM）. The principle is that recording the signal data by using microphone and accelerator. Then, it used the FFT analyzer by fast Fourier transform to execute the signal analysis. To distinguish perfect and cracked eggs, it was found by comparing the resonant frequency and amplitude used microphone and accelerometer as the sensor, Secondly, the results was verified by Support Vector Machine both of method are successful. However, the microphone sensor is better. The results showed that the characteristic frequency of the perfect egg was 4130 ~ 5500Hz and its amplitude was 0.16 ~ 0.20V.However, the spectrum of the cracked egg was messy with no obvious characteristic frequency, and the maximum amplitude was 0.06V. This feature was judged by SVM, and the identification accuracy can reaches to 99% and 98% for training set and the testing set. If that is not knock on the paddy or lump of soft soil on the eggshell, Microphones are used in dirty eggs and crack eggs, and the results of SVM accuracy could reaches to 100% and 100% for training set and the testing set. The conclusion is that resonance detection method to determine the cracks of eggshells is an effective method.