Application Of Low-field NMR And Chemometrics In Rapid Detection Of Leisure Dried Tofu

With the rapid development of China's bean products industry,leisure dried tofu is becoming more and more popular with consumers because of its delicious,nutritious and convenient consumption.However,the quality and safety of leisure dried tofu has become a prominent problem.Accordingly,a fast,objective and effective method is urgently needed by modern bean products enterprises and relevant regulatory authorities for detection of leisure dried tofu.Low field nuclear magnetic resonance spectrometer(LF-NMR)has the characteristics of rapid detection,environmental protection,and simple operation,and has become an important means of detecting food.This article uses LF-NMR to quickly detect leisure dried tofu,and combines chemometrics methods to process the results.In this paper,LF-NMR was applied to the rapid detection of leisure dried tofu,and the chemometrics method was used to process the detection results in order to explore new methods for the detection of leisure dried tofu,and This provides the relevant theoretical basis and experimental basis for application of LF-NMR in detecting leisure dried tofu.The main results of the research are showed below:1.LF-NMR parameter optimizationThe parameters of the Carr-Purcell-Meiboom-Gill(CPMG)pulse sequence of LF-NMR were optimized by using leisure dried tofuas the measurement object.The CPMG sequence parameter for measuring leisure dried tofu is:The cyclic delay(RD)is Is,P90 is 5.45 ? s and P180 is 10.9? s,The 90° and 180° pulse intervals(TAU)is 3000 ? s,The frequency sweep width(SF)is 23.4MHz,Probe dead time(Deadl)is 10?s,Echo number(NECH)is 256,Each scan is 16,Gain multiple(RG)is 40.98.2.Rapid discrimination of the brand of leisure dried tofu based on LF-NMR.Selecting 4 kinds of common leisure dried tofu brand as the research object,and the samples were measured by using LF-NMR.The experimental data were analyzed by pattern recognition methods of principal component analysis(PCA),partial least squares-discriminant analysis(PLS-DA)and Bayesian regularization back-propagation artificial neural network(BR-BP-ANN).The research purpose is to search for a method which can quickly discriminate the brand of leisure dried tofu.The experimental results show that each brand was difficult to be picked out with eyes by three-dimensional PCA scores plot.Further,measurement data of dried tofu were treated by PLS-DA.The results displayed that the total recognition rate of all brands was 86.3%for the training set.For the prediction set,the recognition rate of the brand was 81.3%.However,the BR-BP-ANN method can discriminate four bands simultaneously.The predicted value is in good agreement with the actual expected value for training set.It's interesting to note that the correct rate of prediction was 100%for brand discrimination.the BR-BP-ANN can be a good realization of the identification of dried bean brand.3.Rapid identification of true or false brand of leisure dried tofu based on LF-NMRSelecting a number of common leisure dried tofu brand and "three no"brand as the research object,and the samples were measured for obtaining transverse relaxation data by using LF-NMR.The experimental data were analyzed by pattern recognition methods of PCA,PLS-DA and BR-BP-ANN.The research purpose is to search for a method which can quickly discriminate the brand of leisure dried tofu.The experimental results show that the true and false brands were difficult to be picked out with eyes by three-dimensional PCA scores plot.Further,measurement data of dried tofu were treated by PLS-DA.The results displays that the recognition rate of the true brand was 65.2%,and the false brand was 92.9%for the prediction set.However,the BR-BP-ANN method can obtain better results than PLS-DA.The recognition rate of the true brand was 97.8%,and the false brand was 98.8%for the prediction set.4.Rapid determination of moisture content in leisure dried tofu based on LF-NMRThe samples of leisure dried tofu were measured by using LF-NMR for obtaining transverse relaxation data.Direct drying method was used to determine moisture content of dried tofu.The measured results were used as chemical values.The model was established by employing the partial least squares(PLS)and back-propagation artificial neural network(BP-ANN)methods combined with NMR data and chemical values.It realized the rapid determination of moisture content of dried tofu.The experimental results that the calibration and prediction moisture content models respectively gave the correlation coefficients of 0.9235 and 0.9189 for PLS and of 0.9176 and 0.9215 for BP-ANN;the root mean standard errors of calibration(RMSEC)were 0.0272 for PLS and 0.0281 for BP-ANN,respectively;the root mean standard errors of prediction(RMSEP)were 0.0248 for PLS and 0.0223 for BP-ANN,respectively.The study showed that the two methods can quickly and accurately predict the moisture content of leisure dried tofu.