| Mandarin fish has firm meat,delicious taste and rich nutrition.It is a special economic freshwater fish in China.Umami intensity is one of the most important quality indicator of mandarin fish.Existing umami taste evaluation methods normally require damage of samples,expensive or cumbersome procedures,etc.Therefore,it is necessary to develop a fast,non-destructive,accurate,and inexpensive umami taste intensity detection method.In this paper,the pure solution of sodium glutamate(MSG),the mixed system of MSG and disodium inosinate(IMP),and the mixed system of MSG,IMP and other umami substances as well as mandarin fish were used as the research objects,and near-infrared spectroscopy was combined with umami taste evaluation method to quantitatively detect of umami taste substances and umami taste intensity of mandarin fish.The main research conclusions were as follows:(1)The feasibility of using near-infrared spectroscopy(NIRS)to detect the concentration of umami substances was verified.By comparing different pretreatments and different feature wavelength selection methods,the full wavelength and simplified models for detecting the concentration of umami substances in single and binary mixed systems were established.The research results showed that both the full-wavelength and simplified models yielded excellent predictions of the concentration of substances in the two systems,and the optimal model prediction correlation coefficients of each modeling method were all above 0.90.Among them,the highest prediction accuracy was attained by the simplified USS-SPA-PLSR model of MSG concentration for the pure solution,where the number of wavelengths was 35,Rp=0.9888?RMSEp=0.2063g/L.(2)The feasibility of using NIRS to detect the umami taste intensity of a mixed system of different umami taste substances was verified.Different pretreatment and feature wavelength selection methods were used to establish the full wavelength and simplified models for the evaluation of the umami intensity of the binary hybrid system EUC and the ternary hybrid system.It was demonstrated that the optimal model prediction correlation coefficient of each modeling method for the binary hybrid system EUC detection was greater than 0.85,and the prediction effect was good.The simplified USS-SPA-ELM model with the highest prediction accuracy was the simplified USS-SPA-ELM model with a wavelength of 23,Rp=0.9335?RMSEp=2.772g/dL.The prediction correlation coefficients of the ternary mixed system umami intensity detection optimal models were all greater than 0.80.After the feature wavelengths were selected,the model was greatly simplified,while the prediction accuracy is greatly improved,and the optimal prediction correlation coefficients were all above 0.90.The model with the highest prediction accuracy is the simplified USS-CARS-BP model of the umami intensity of the ternary system of MSG,IMP and TMAO.The number of wavelengths was 18,Rp=0.9762?RMSEp=0.7452.(3)The full wavelength and simplified quantitative detection model of mandarin fish umami substance content and umami intensity intensity were determined.The results of the study showed that NIRS was promising in determining hypoxanthine nucleotide(IMP),adenine nucleotides(AMP),guanine nucleotides(GMP),aspartic acid(Asp)and glutamic acid(Glu)and the flavor of mandarin fish Intensity has better detection effect,and the correlation coefficients of the optimal model predictions for each index are all between 0.80 and 0.94.In the detection of umami substances,the highest prediction accuracy is the simplified USS-UVE-BP model of Asp concentration,with a variable number of 31,Rp=0.9347?RMSEp=0.2327mg/100g.In the detection of umami intensity,the highest prediction accuracy is the simplified USS-UVE-ELM model,the number of variables is 32,Rp=0.8772?RMSEp=1.322.The results of this study show that it is feasible to use near-infrared spectroscopy to detect the concentration and intensity of umami-flavor substances in solutions and mandarin fish. |
PDF Full Text Request