Prediction of Egg's Freshness Using Backward Propagation Neural Network

Abstract

Chemical, biological, and physical changes occur in eggs with the increase in the number of storage days and the storage condition. One hundred twenty eggs laid on the same day ranging in color from light brown to dark brown and of different size were bought from the local market. The samples were stored in the lab room ambient temperature and relative humidity. The spectral data from 300 to 1100 nm of 20 sample eggs on each alternative days were obtained until the 12th day of storage. Predictive model for Haugh Unit (HU) and thick albumen height was developed using backward propagation neural network (BPN) with Savitzky Golay (SG) third pass filter and Multiplicative Scatter Correction (MSC) pre-processing method. BPN predicted HU with correlation coefficient (R) and Root Mean Square Error of Calibration (RMSEC) of 0.82 and 8.11, respectively. Similarly the BPN predicted the thick albumen height with R and RMSEC of 0.83 and 0.88 mm. The device used in the research for spectral collection is simple and portable and the results obtained are satisfactory, this method can be used to evaluate the freshness of eggs in terms of HU and thick albumen height in the local market.