Detection Of Benzoyl Peroxide Additive Amount In Flour By Near Infrared Spectroscopy

Flour is one of the essential ingredients in people’s daily dietary structure. Addingbenzoyl peroxide (BPO) can improve the flour apparent quality. But the excessiveBPO dopants will bring hidden troubles to consumers’ health.The purpose of thispaper was to study and bring forward a method for quick detection of BPO additiveamount in flour. This paper use the near-infrared spectroscopy for the detection ofBPO addition, the main contents are organized as the following:1、This paper gives an analytical method by spectrometry of “equivalentdetection” on BPO in the flour. The feasibility of “equivalent detection” wasillustrated by the theories of spectral analysis and verified with the results ofexperiment. Meanwhile, the effect of different spectral pretreatment methods ofwavelet decomposition (WD) and empirical mode decomposition (EMD) upon themodeling accuracy of BPO addition in flour was explored as well.2、Effects of temperature on the measurement results in flour were studied. Theresults show that: the stability of model for predicting BPO addition was affected bythe changing with temperature, it presented as the model developed at onetemperature used to predict the BPO addition at other temperature has a lowforecasting precision. This paper discusses several different methods in an attempt tocompensate for temperature effects. These include, the mixed temperature correctionmodel can improve the temperature adaptation of the model; the direct orthogonalsignal correction (DOSC) method used for correct the spectral data correction canimprove the model accuracy, but the temperature adaptation of the model is stillunsatisfying; the direct standardization (DS) and piecewise direct standardization(PDS) method used for spectral data correction can be better to eliminate the impactof temperature on the test results, and improve the stability of the predict model atdifferent temperatures.3、This paper presents a pre-treatment method to concentrate the extraction of theflour sample to improve the detection accuracy of the BPO addition. It means BPOand its reduction product BA in the flour samples extracted by organic solvent ofmethanol; then the extract of each sample was concentrated more than10fold bynitrogen blow-down method; the spectra of the concentrate liquid was collected andmodeled with the BPO addition of the test samples, the results show that the methodcan improve the detection accuracy of BPO addition effectively. 4、The best optical path length will help to improve the measurement accuracy ofBPO addition when analyze the concentrate liquid of the flour samples byspectrometry method. However, it is difficult to obtain the exact best optical pathlength in actual measurement. So an analysis model of using spectra from thecombination of multiple optical path lengths and the concentrate liquid was used inthis paper: The spectral modeling method of the combination of multiple optical pathlengths can further improve the detection accuracy of BPO addition, but it alsoincrease the amount of spectral data and raise the complexity of the model. In thissituation, a novel modeling analysis methods of the combination of sensitivewavelengths from multi path lengths were promoted in this paper. That is firstlyextracting several sensitive wavelengths by the wavelength selection methods ofgenetic algorithm and successive projections algorithm, etc; and then using thecombination spectra of sensitive wavelengths for modeling analysis, the resultsshowed that: the modeling analysis methods of the combination of sensitivewavelengths from multi path lengths can substantially reduce the amount of spectraldata and the complexity of the model but keep modeling accuracy at the same time.