The Application Of Electronic Tongue In Detection Of Vinegar Quality And Monitoring Of Vinegar Fermentation

Vinegar is a traditional condiment in China,which has anti-fatigue,anti-oxidation and some other health-care function.With the rapid development of the food industry, the types of vinegar are increasingly rich.Otherwise,there are many food safety problems with this.The main problems of vinegar quality were found,such as total acid content of standard,benzoic acid,total number of colonies,cyclamate and saccharin sodium exceeded standard.For the huge profits,some manufacturers produce false vinegar mixed with industrial acetic acid which have hazard effect on human health with long-term consumption.In order to meet consumers' demands for the quality of vinegar,it is necessary to develop a method which could determine the vinegar quality rapidly and monitor the fermentation procedure.Electronic tongue which uses sensor arrays instead of taste buds cells to detect biological liquid sample, is an new type of modern instrumentation which simulates the mechanism of human taste.And patter recongnition methods were used to the results.The output of electronic tongue is not the result of sample composition,but some characteristics of the signal mode which showed the overall evaluation of taste characteristics. Compared to the method of the human sensory evaluation and chemical analysis methods,electronic tongue has good reproducibility,rapid measurement and simple operation.The main contents of this study was using electronic tongue to detect vinegar of different brands and different processes,and the data obtained by electronic tongue was analyzed by different pattern recognition methods,such as discriminant analysis, K nearest neighbor method and BP neural network.The effects on detection results of vinegar by different cleaning solution were also researched.Electronic tongue was used to monitor vinegar fermentation by detecting key compounds of fermentation, such as total acid,fixed acid,reducing sugar and amino acid nitrogen,and the data was analyzed by multiple linear regression,partial least squares and BP neural network to establish the relativity with the results of electronic tongue and chemistry components.The main conclusions are presented as following: 1.Different kinds of vinegar in different concentration were detected by electronic tongue and the data was analyzed by Fisher linear discrimination.The result of the alternation validate recognition rate of mature vinegar,aromatic vinegar and white vinegar was all 100%,while the rice vinegar is 96.7%.The influence of different cleaning solution to division results was discussed and display quite. Different optimum cleaning solvent:were chosen with respect to different vinegar: mature vinegar and rice vinegar was 4%acetic acid,aromatic vinegar was 3%acetic acid and white vinegar was clean water.Different craftwork vinegars were detected using electronic tongue,4%acetic acid was considered to be the optimum cleaning solvent by variance analysis.The prediction recognition rate was 100%,used by KNN and BPANN.The model had great preciseness,which could discriminate vinegar quality rapidly through detecting vinegar sense with electronic tongue.2.Electronic tongue was used to monitor aromatic vinegar fermentation procedure while and Zhenjiang Hengshun aromatic vinegar was selected as the research object.The change of total acid,fixed acid,reducing sugar and amino acid nitrogen were detected by electronic tongue and chemistry method respectively. Different patter recognition methods,such as MLR,PLS and BPANN were compared. The conclusions were drawn as follows:BPANN was the best imitation method to fit for total acid,fixed acid and reducing sugar.The determinant coefficients(R~2) based on calibrations of BPANN,were 0.8439,0.9382,0.8322.respectively.PLS was the best imitation method to amino acid nitrogen and the R~2 was 0.8751.MLR was the worst compared with other methods.Root mean standard errors of prediction (RMSEP) of total acid,fixed acid,reducing sugar and amino acid nitrogen were 0.8241,0.0963,0.5557 and 0.1482,respectively.Since the relationship between the sensor response value and ingredients concentration was mostly non-linear,BP-ANN model develop which has non-linear mapping function was found to have better results and prediction accuracy than PLS and MLR.This research indicates that the electronic tongue could be used for quantitative analysis in ferment production.It is feasible to monitor fermentation procedure of vinegar using the method developed.