Abstract: In this study, an electronic nose was used to detect the flavor of Cantonese sausage during processing (0, 12, 24, 36, 42, 48, 54, 60, 72, 96 and 120 h) and storage (0, 2, 4, 6, 8, 12 and 20 weeks). The acid value (AV) and peroxide value (POV) were simultaneously measured to evaluate fat oxidation in Cantonese sausage. The contribution rates of 10 sensors to the flavor of sausages were evaluated through loading analysis, variance analysis and Pearson analysis. The results indicated that as the optimal sensor arrays for monitoring fat oxidation, S4, S6, S7, S8 and S9 for processing and S6, S7, S8 and S9 for storage were selected. Artificial neural network (ANN) models were developed to predict the degree of fat oxidation in Cantonese sausage using electronic nose data. The R2 values of the models based on all sensors for AV and POV prediction during processing were 0.959 and 0.930, respectively, while those were 0.930 and 0.914 based on the optimal sensors, respectively. During storage, all the R2 values were greater than 0.9, except for the POV prediction model based on the optimal sensors with R2 value of 0.805. Therefore, the electronic nose was suitable for evaluating fat oxidation in Cantonese sausages during processing and storage, which could be further applied to guide the commercial production and storage of Cantonese sausage.

Key words: Cantonese sausage, electronic nose, fat oxidation, artificial neural network, prediction model