Rapid Detection Of Benzo(a)pyrene And Peroxide Value In Peanut Oil Based On Raman Spectroscopy And Electronic Nose

Peanut oil is essential in people’s daily diet.Its quality and safety are related to industrial development and national health.Traditional peanut oil quality and safety detection methods are often time-consuming and laborious.It is difficult to meet the demand for real-time monitoring,so a rapid detection method is urgently needed.In this paper,the advantages of wide detection range,high sensitivity and aqueous solution measurement of Raman spectroscopy were used.Combined with electronic nose technology,machine learning algorithms were used to carry out research on the rapid detection of peanut oil quality and safety.The main findings were as follows.（1）A rapid detection method for benzo[a]pyrene concentration in peanut oil based on Raman spectroscopy combined with machine learning methods was developed.The effects of glass substrates and magnetron sputtered gold substrates on the Raman spectra of peanut oil were compared.The high-dimensional Raman data was pretreated using principal component analysis（PCA）and t-distributed Stochastic Neighbor Embedding（t-SNE）methods.Qualitative and quantitative detection models for benzo[a]pyrene concentrations in peanut oil were developed using back propagation neural network（BPNN）,partial least squares regression（PLSR）,support vector machine（SVM）and random forest（RF）algorithms.The detection limit of Raman spectroscopy of benzo[a]pyrene concentrations in peanut oil were validated.The results showed that the Raman spectra of glass substrates were more suitable for the detection of benzo[a]pyrene than magnetron sputtered gold substrates.The accuracy of the pretreated Raman data combined with machine learning methods in the concentration level identification models reached 100%.The correlation coefficient（R）,the root mean square error（RMSE）and the Bias of the prediction set in the quantitative model were0.9996,0.1853μg/kg and 0.0121μg/kg,respectively.（2）A combination of surface-enhanced Raman spectroscopy（SERS）and electronic nose technique（E-nose）was used to achieve rapid and accurate detection of peanut oil peroxide value.The best SERS spectral region associated with peanut oil oxidation was found to be 799 cm^-1～1073 cm^-1 using the i PLS method.A sensor array was designed for the characteristic components of peanut oil peroxide gas.The average response value and stable response value of the response curve of the gas sensor array were screened as the characteristic values.Three different levels of data fusion strategies were compared.Qualitative identification and quantitative prediction model of peanut oil peroxide value based on SERS and electronic nose information was established and compared with the performance of single data mode.The identification analysis of peanut oil oxidation status was also performed.The results showed that the data fusion strategys could effectively improve the model prediction performance.PCA-SVM could accurately identify the oxidation state of peanut oil.In the low-level fusion strategy,BPNN and RF well identified the peroxide concentration of peanut oil,both A_c and A_p reached 100%.In the med-level fusion strategy,t-SNE-RF model obtained the best quantitative prediction of peroxide value,with R_c,R_p,RMSEC,RMSEP and Bias were 0.9991,0.9987,0035 g/100g,0.047 g/100g and 0.0014 g/100g,respectively.The results of this study showed that the rapid detection of peanut oil benzo（a）pyrene and peroxide value could be achieved by using Raman spectroscopy and electronic nose techniques in combination with appropriate chemometric methods.The development of this study provided a theoretical basis for the application of rapid detection of edible vegetable oil quality and safety based on Raman spectroscopy and electronic nose techniques,and also explored new methods for the rapid detection of other food products.