Premium
Application of Visible/Near‐Infrared Spectroscopy in the Prediction of Azodicarbonamide in Wheat Flour
Author(s) -
Che Wenkai,
Sun Laijun,
Zhang Qian,
Zhang Dan,
Ye Dandan,
Tan Wenyi,
Wang Lekai,
Dai Changjun
Publication year - 2017
Publication title -
journal of food science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.772
H-Index - 150
eISSN - 1750-3841
pISSN - 0022-1147
DOI - 10.1111/1750-3841.13859
Subject(s) - partial least squares regression , wheat flour , mathematics , mahalanobis distance , correlation coefficient , dilution , near infrared spectroscopy , artificial neural network , mean squared error , statistics , pattern recognition (psychology) , computer science , artificial intelligence , food science , chemistry , physics , optics , thermodynamics
Azodicarbonamide is wildly used in flour industry as a flour gluten fortifier in many countries, but it was proved by some researches to be dangerous or unhealthy for people and not suitable to be added in flour. Applying a rapid, convenient, and noninvasive technique in food analytical procedure for the safety inspection has become an urgent need. This paper used Vis/NIR reflectance spectroscopy analysis technology, which is based on the physical property analysis to predict the concentration of azodicarbonamide in flour. Spectral data in range from 400 to 2498 nm were obtained by scanning 101 samples which were prepared using the stepwise dilution method. Furthermore, the combination of leave‐one‐out cross‐validation and Mahalanobis distance method was used to eliminate abnormal spectral data, and correlation coefficient method was used to choose characteristic wavebands. Partial least squares, back propagation neural network, and radial basis function were used to establish prediction model separately. By comparing the prediction results between 3 models, the radial basis function model has the best prediction results whose correlation coefficients (R), root mean square error of prediction (RMSEP), and ratio of performance to deviation (RPD) reached 0.99996, 0.5467, and 116.5858, respectively. Practical Application Azodicarbonamide has been banned or limited in many countries. This paper proposes a method to predict azodicarbonamide concentrate in wheat flour, which will be used for a rapid, convenient, and noninvasive detection device.