Detection of submerged fermentation of Tremella aurantialba using data fusion of electronic nose and tongue

Abstract

AbstractIn this study, a fusion model to combine electronic nose and electronic tongue was developed to detect submerged fermentation of Tremella aurantialba (T. aurantialba). Three chemical indicators of T. aurantialba fermentation were detected by traditional measuring methods as a comparison, including ergosterol content, reducing sugar content and polysaccharide content. Principal component analysis (PCA) and layer fusion in the feature were applied on electronic nose and electronic tongue data. Support vector machine regression (SVR) model was established for quantitative determination of the three chemical indicators. The SVR model was combined with PCA for testing of ergosterol content, reducing sugar content, and polysaccharide content. The correlation coefficient of the test set for ergosterol content, reducing sugar content, and polysaccharide content were 0.9946, 0.9945, and 0.9934, respectively. The root mean square error of prediction of the models for the three indicators were 2.7765 μg/mL, 29.1405 mg/mL, and 0.4858 mg/mL, respectively. The results proved that the combined system could be used to predict the ergosterol content, reducing sugar content, and polysaccharide content during submerged fermentation of T. aurantialba. In conclusion, detection of T. aurantialba fermentation based on electronic nose and electronic tongue technologies achieved satisfactory results and, therefore, offers broad application prospects in the liquid state industry.Practical applicationsThe fermentation broth of T. aurantialba has a variety of pharmacological effects. The submerged fermentation of T. aurantialba is very complex. The existing detection technology needs a cumbersome process of sample preparation and costs a lot of time, which cannot meet the need of rapid detection. Electronic nose combined with electronic tongue could detect the odor and taste changes during the T. aurantialba fermentation quickly. The fusion techniques showed good performances in the quantitation of the main indicators of T. aurantialba. The method can be used to determine the state of fermentation instead of the traditional methods. It further improves the accuracy and scientificity of quantitative evaluation for T. aurantialba fermentation. Data fusion technique to combine electronic nose and electronic tongue could be applied in the liquid state fermentation industry in the future.