Abstract: This research was done to improve the objectivity of quality evaluation of dried aquatic products and to establish a reliable evaluation model based on electronic nose. Salted silver carp dried under different drying conditions by a heat-pump dryer were detected by electronic nose and measured for total volatile basic nitrogen (TVB-N). Principal component analysis (PCA) combined with statistical pattern recognition and partial least squares (PLS) regression analysis was used to identify TVB-N values of dried silver carp. The results showed the electronic nose had different responses to the TVB-N values of the silver carp dried under different conditions. The recognition rates of training set and testing set based on the K nearest neighbor (KNN) analysis model were over 90%. The correlation coefficients of the PLS model for training set (containing 30 samples) and testing set (containing 15 samples) were 0.913 5 and 0.933, respectively. The root mean square error of cross validation (RMSECV) was 3.67, and the root mean square error of prediction (RMSEP) was 3.26. Results showed that the predictions of the PLS model based on TVB-N could provide greater accuracy in identifying fish dried under different conditions. In conclusion, as an intelligent instrument for the analysis and identification of olfactory signals, electronic nose is a potential objective tool for evaluating the quality of dried fish.

Key words: dried salted silver carp, quality evaluation, electronic nose, model, heat pump drying