Study Separating Technology On Egg By Near Infrared Spectroscopy

Eggs are an important high-protein food on the table all over the world.They are loved for their affordable price,good taste and many ways of eating.But because the production of eggs is very large,and easy to be damaged,so the large-scale non-destructive testing of eggs is the current development of the egg industry is very urgent requirements.The nondestructive classification of eggs can be carried out by near infrared spectroscopy.In this paper,nir(near infrared spectroscopy)nondestructive testing technology was applied to the separation of egg varieties.The main research contents and results are as follows:1.The identification and sorting models of normal eggs and damaged eggs,caged eggs and free-range eggs,white eggs and brown eggs were established.In the same environment of spectral information collection,the near-infrared spectral information of eggs was collected,and the differences in spectral information of different types of eggs were analyzed.Three different pretreatment methods(first derivative,second derivative,multiple scattering correction)were used to select characteristic bands representing different egg species.2.On behalf of the eggs will filter out the characteristics of the band by the type of input the BP neural network and RBF neural network,through the correlation coefficient 2,mean square error(MSE discriminant model and predicted variance RMSEP identification ability.The study shows that the identification effect of BP neural network model is better than that of RBF neural network model.After multiple scattering correction of the original spectral data to identify the best effect,the correlation of 2 = 0.999,the mean square error(MSE and prediction mean square error RMSEP are tend to be 0,can be achieved for egg kind of sorting.