Prediction Model for Soluble Solids Content of Apple——Based on Visible/Near Infrared Spectroscopy Technology

Zhou Liping,Hu Yaohua,Chen Da,Guo Kangquan,Yue Tianli
DOI: https://doi.org/10.3969/j.issn.1003-188X.2009.04.032
2009-01-01
Abstract:To predict the soluble solid content(SSC) of apples quickly,this paper reports on a new prediction model for apple's SSC by means of visible/near infrared spectroscopy(Vis/NIRS)(345~1039).The reflectance spectra were acquired by using the spectroscopy.Principal component analysis was used to analyze the characteristics of the spectra.Five new principal components were confirmed to describe the whole body of spectra when the total reliabilities arrived at 95%.Then the five new principal components were applied as the input of back propagation(BP) neural network with one hidden layer.The values of SSC were applied as the output of BP neural network.The results showed that the relative error of over 98% prediction samples was under 5% and this prediction model was reliable and practicable.
What problem does this paper attempt to address?