Electronic Tongue-Based Prediction Models Predicting Physicochemical Parameters of Chinese Herring Solid-State Fermentation
Article
Figures
Metrics
Preview PDF
Reference
Related
Cited by
Materials
Abstract:
To monitor the process of solid-state fermentation of Chinese herring, parameters including water content, pH, total acid content, amino nitrogen (ANN) content, total volatile basic nitrogen (TVB-N) content, and changes in the taste fingerprint were measured by conventional analytical methods and an electronic tongue. The principal component analysis (PCA) and discriminant analysis (DA) were used to identify fermented Chinese herring samples undergoing different durations of fermentation, while partial least-squares regression (PLSR) prediction models for the electric tongue and related physicochemical indicators were also established and evaluated. The results indicated that the physical indicators and taste of Chinese herring samples changed significantly during fermentation. The accumulative contribution rate of three extracted principal components was 94.49%, the discrimination coincidence rate was 100%, and the Chinese herring samples with different fermentation durations could be identified effectively by both, PAC and DA. In five prediction models based on electronic tongue signals, both relative percent deviation (RPD) values of water content and ANN models were 1.80, and the RPD value of TVB-N model was 2.47, which could be used for qualitative analysis. The RPD values of pH and total acids models were both > 5, which indicated good quantitative effect, good stability, and high prediction accuracy. Thus, it is feasible to identify and monitor the process of solid-state fermentation of Chinese herring using the electric tongue coupled with related chemometric methods.
