Eggs are widely consumed in many products industry and homes as they are rich in vitamins and minerals. In order to meet increasing need quickly, automation has been made in chicken farms for processes, such as collecting eggs, weight classifying, separating cracked, and packing. If shell is cracked, harmful microorganisms can easily enter into it, and egg will deteriorate in a short time due to contact with air. Cracks can be large enough to be visible to naked eye, and sometimes they are micro-sized and cannot be detected by human eye. In this study, detection of cracked eggshell based on signal processing and machine learning was carried out. Acoustic signal generated as a result of impact made to shell by means of mechanical system was recorded for 0.2 seconds at a sampling frequency of 50kHz with microphone. Separately, 50 eggs data with intact and cracks shells were recorded with system and data set were created. Threshold value of 0.74V was used to determine time from moment of impact to egg shell to damping, and 680 data were taken after this value. The detail and approximation components with different frequencies were extracted by applying Wavelet Packet Transform (WPT) from 2nd level with db4 main wavelet. By calculating entropy value of each component, 1x4 feature vector was obtained. Artificial Neural Network (ANN) was used to determine efficiency of extracted feature vector in detecting crack egg shell. 100% performance was achieved and an egg's shell crack detection time was determined in approximately 0.216 seconds.
: December 27, 2020
|APA||Balcı, Z , Yumurtacı, M , Yabanova, İ , Ergin, S . (2021). Yumurta Kabuğundan Alınan Akustik Sinyalin Dalgacık Paket Dönüşümü ve Entropiye Dayalı Olarak İşlenmesi ve Yapay Sinir Ağlarıyla Çatlağın Belirlenmesi . Bilecik Şeyh Edebali Üniversitesi Fen Bilimleri Dergisi , 8 (1) , 125-135 . DOI: 10.35193/bseufbd.847763|