Ultraviolet Spectrum Characteristic Analysis And Quantitative Detection Of Escherichia Coli In Milk

Rapid and accurate detection of total food-borne pathogenic microorganisms is an important prerequisite for real-time monitoring of total microorganisms in food processing.Spectroscopic detection has broad prospects in the field of microbial detection because of its advantages of fast,non-destructive and simple operation.However,due to the complex growth environment of microorganisms,the red shift of characteristic peaks,overlap of peaks among components and obvious spectral noise are easy to occur in the detection spectra of microorganisms.Although we can pick up the characteristics of the spectrum in one-dimensional spectral signals,but it is difficult to accurately isolate the characteristic wavelengths that characterize the growth of microorganisms.resulting in low accuracy of the total number of detection results.Therefore,this paper applies two-dimensional correlation analysis method to the study of microbial spectral characteristics picking up and analysis in complex environment,and takes milk as the detection background,and E.coli as the detection object,which provides theoretical basis for further rapid and accurate detection of microbial total in complex environment.Aiming at the red shift of spectral characteristics of E.coli caused by the complex composition of milk.Firstly,the ultraviolet spectrum of E.coli in milk under continuous culture time was obtained,and the distribution of characteristic peaks in red shift spectrum was analyzed from the point of view of the types and quantities of conjugated amino acid molecules in E.coli and the solvent effect caused by milk,and the characteristic bands were determined.In view of the overlap of characteristic spectra between milk components and E.coli.Based on the advantages of two-dimensional correlation analysis in spectral line separation,In this paper,two-dimensional correlation analysis of characteristic band spectra was carried out to separate the overlapping spectra of milk and E.coli.According to Noda rule,the characteristic wavelengths of two-dimensional spectra which can characterize the viability of E.coli were picked up.The model relationship between the characteristic wavelength corresponding absorbance value and the total number of E.coli in milk was established by partial least squares method,and compared with one-dimensional spectral pickup characteristic wavelength model,In order to verify the feasibility of two-dimensional correlation analysis in accurately picking up the spectral characteristics of E.coli.Baseline drift and spectral noise interference caused by light scattering of proteins and fats in milk were studied.The effect of baseline correction and twodimensional atlas denoising on the accuracy of E.coli spectral feature pickup was studied to optimize the prediction model.The results show that the S-G convolution smoothing method based on SNV and two-dimensional operation has the best optimization effect on the original model,and the correlation between the predicted value and the real value is more than 95%.