| Rapeseed meal is a by-product of rapeseed oil extraction,with high protein content but low bioavailability.It was found that rapeseed meal can degrade macromolecular protein into polypeptides by solid state fermentation,which can effectively improve the utilization of rapeseed protein and greatly increase its value.However,in the solid-state fermentation process,the traditional offline measurement method used to detect the fermentation parameters was time-consuming and laborious,which seriously impeded the improvement of the solid-state fermentation process control and fermentation optimization.Near infrared spectroscopy has many advantages,such as fast detection,nondestructive and nonpolluting.This technology showed great potential in the rapid detection of parameters in the solid-state fermentation process.Thus,this paper explored the application of near-infrared spectroscopy in the preparation of peptides from solid-state fermented rapeseed meal by Bacillus subtilis,to achieve rapid detection of key parameters of fermentation process.The main research contents and results are as follows:(1)Rapid discrimination of rapeseed meal in solid state fermentation stage was realized by near infrared spectroscopy.The fermentation process indexes,such as matrix moisture content,pH,color difference and microbial content,were determined by traditional test method.These indexes data and the fermentation time data were treated with MATLAB software for clustering analysis,to get the result that the meal solid-state fermentation could be divided into four stages: initial stage(0~5 h),biomass rapid growth(5~12 h),accelerate fermentation period(12~26 h)and stable fermentation period(26~72 h).The above mentioned samples for indexes measurement were also scanned by near-infrared spectroscopy simultaneously.Standard normal variate(SNV),multivariate scatter correlation(MSC),savitzky-golay(S-G),Normalization and derivative correction were employed for spectral data preprocessing.K-nearest neighbors(KNN)was used for pattern recognition,the preprocessing method,principal component number and K value were cross-validated.And after the first derivative preprocessing,under the condition that the principal component is 15,and K is 1,the model prediction result is optimal,the discrimination rate of 126 training set samples was 100%,and the recognition rate of 42 prediction sets was 95.24%.The results show that the method can effectively and quickly discriminate the solid-state fermentation stage of rapeseed meal,and provide technical basis for the stage regulation of fermentation process.The results showed that this method could effectively and rapidly discriminate rapeseed meal in the solid-state fermentation stage and provide a technical basis for the stage regulation of the fermentation process.(2)Near infrared spectroscopy model of microbial biomass in fermentation process was established.The fermentation samples from solid-state fermentation of rapeseed meal were collected,the microbial biomass was determined by plate counting method and the near-infrared spectrum was obtained simultaneously by scanning the fermentation samples.The partial least squares(siPLS),interval partial least squares(iPLS)and synergy interval partial least squares(siPLS)were used to establish the quantitative microbial model.The results showed that the joint interval partial least square model was optimal,and its pretreatment method was Normalization.The optimal spectral wavelength range was 1,135.2~1,188.8 nm,1,388.0~1469.0 nm,1,666.3~1,784.4 nm,and 1,785.6~1,922.0 nm.The rc and RMSECV of the model calibration set were 0.9402 and 0.639,respectively.The rp and RMSEP of the model prediction set were 0.9521 and 0.618,respectively.The results showed that the detection of microbial biomass in solid-state fermentation of rapeseed meal could be realized by near infrared spectroscopy with low error.(3)A near infrared spectroscopy model for the content of polypeptides in fermentation matrix was established.Spectral scanning and polypeptide content determination were performed on 128 samples.The obtained near-infrared spectrum was preprocessed,and then cross-validated with three modeling methods of PLS,iPLS and siPLS.The results showed that the siPLS model had the best prediction effect,and the spectral pretreatment method was the combination of MSC and Normalization.The optimal spectral intervals were screened as 1,135.2~1,174.8 nm,1,175.3~1,217.8 nm,2,041.5~2,173.3 nm,and 2,327.4~2,500.2 nm.The polypeptide model training set rc was 0.9401,RMSECV was 2.09,rp was 0.9284 and RMSEP was 2.4.The results showed that the model had good correlation and small error,and could be used to predict the content of polypeptides in solid-state fermentation of rapeseed meal.(4)The transfer of pH quantitative models between different instruments in the solid-state fermentation process was studied.The pH value changed during solid-state fermentation were measured and the near-infrared spectra were scanned by both the source instrument(integrating sphere module)and the new instrument(ray probe module).The pH quantitative model of source instrument spectrum was established by siPLS,and the calibration set rc was 0.9716,RMSECV was 0.172,the prediction set rp was 0.9616,and RMSEP was 0.211 in the optimal siPLS model.Moreover,the effects of four model transfer methods,S/B,DS,PDS and Shenk’s,were compared respectively,the results showed that the source spectral model had the best prediction effect after being transferred by Shenk’s algorithm,the correlation coefficient rp of the prediction set was increased from 0.7034 before transformation to 0.9563,and the standard root-mean-square error RMSEP reduced from 1.90 before transformation to 0.269.The results showed that the pH quantitative model in the solid-state fermentation of rapeseed meal could realize the transfer between different near-infrared instruments,improve the versatility of the model,save the cost of repeating modeling.This provide an effective way for the model transfer of other key parameters in solid-state fermentation. |
PDF Full Text Request