Monitoring The L-isoleucine Fermentation Process By Near-infrared Spectroscopy

In the process of L-isoleucine fermentation using Brevibacterium flavum KM011,real-time understanding of bacterial concentration,glucose concentration and accumulation of major by-products has important significance for the control of the whole fermentation process.This not only allows the carbon skeleton to flow to the target product to the maximum extent,but also improves the utilization rate of the fermentation equipment,thereby saving costs and increasing the conversion rate of sugar to acid.At present,the detection of fermentation broth components is still dominated by traditional chemical analysis methods.These methods have problems such as complicated pre-processing,long time-consuming,high cost,reagent loss,and special equipment,which limits the detection efficiency.Compared with the conventional method,Near-Infrared Spectroscopy?NIS?method has the advantages of fast detection speed,no sample pretreatment,low cost,and is a green environmental protection test method.In this paper,NIS analysis technology was combined with chemometrics method to carry out quantitative of five amino acids in L-Ile fermentation process,and amino acid fermentation data management system was established to improve the utilization of current amino acid fermentation data.The main research contents and conclusions are as follows:?1?During L-Ile fermentation process,105 fermentation broth samples were collected.The concentrations of five amino acids were determined by?HPLC?.And transmission spectra and reflectance spectra of the fermentation broth samples were collected using a Thermo Antaris II near-infrared scanner.The best detection parameters of near-infrared spectroscopy instruments under different acquisition modes were determined by one-way analysis of variance.The optimal parameters of liquid transmission acquisition instruments were 32 times of scanning and resolution of 16cm^-1.The optimal parameters of the suspension reflection scanning instrument were 32 times of drawing and the resolution of 8cm^-1.?2?The transmission spectra of 105 fermentation broth samples were collected under the above optimal parameters,and the obtained spectral data was combined with the chemometric method to establish five amino acid spectral correction models.The results showed that different spectral pretreatment methods have different effects on the extraction of effective spectral information of amino acids,and the best prediction model for establishing was selected from 5400-6300 cm^-1+7300-10000cm^-1.the Rc of L-Ile,L-Ala,L-Glu,L-Thr,L-Leu were 0.981,0.909,0.927,0.957,0.649,the RPD of correction set were 7.3,2.9,3.1,3.6,1.6.External verification results indicated that R_p were 0.997,0.925,0.967,0.977,0.10,and the RPD of verification set were 4.7,2.6,2.5,3.3,1.1.It is indicated that the L-Ile,L-Ala,L-Glu,L-Thr spectral prediction models have better prediction effects,while the L-Leu model has no predictive ability.?3?The reflection spectra of 105 fermentation broth samples were collected under the above optimal instrument parameters.SPSS software was used to calculate the correlation between spectral absorbance and product concentration under different spectral pretreatment,and the optimal band was selected according to the correlation coefficient.The reflection spectrum prediction models of five amino acids were established combining PLS.The R_c of L-Ile,L-Ala,L-Glu,L-Thr,L-Leu were 0.987?0.981?0.986?0.992?0.968,and the RPD of correction set were 7.8?6.8?6.3?6.4and 5.0.External verification results indicated that verification set correlation coefficient R_p>0.98,RPD>3,was basically consistent with the calibration model.Compared with the spectral prediction model built by transmission acquisition,the accuracy of the model built by the reflection acquisition spectrum is greatly improved.?4?Considering the utilization of current amino acid fermentation data,an amino acid fermentation data management system was established.The system framework was built using the B/S browser mode,and the database management system has functions of storing,searching,downloading,and visualizing the data as needed,and displaying these functions.