鱼露发酵过程中挥发性盐基氮和氨基酸态氮质量浓度的预测

Prediction of volatile base nitrogen and amino acid nitrogen in fish sauce fermentation process

研究了鱼露发酵过程中理化指标的变化情况并对发酵过程中产生的挥发性盐基氮(TVB-N)和氨基酸态氮(AN)质量浓度进行预测。通过对不同鱼露发酵条件下的理化指标进行检测,得到130组实验数据,由Person相关性分析可知,在整个发酵过程中,TVB-N和AN质量浓度均与发酵时间和含盐量呈正相关性,与pH呈负相关性,与温度和成曲添加量不相关。将主成分回归、偏最小二乘回归以及BP神经网络模型应用于TVB-N和AN质量浓度的预测,结果表明3种模型均能对鱼露发酵过程的TVB-N和AN质量浓度进行预测。比较3种模型的预测参数(相关系数、均方根误差以及相对分析误差),发现BP模型具有最低的均方根误差与最高的相关系数和相对分析误差,表明BP模型能更准确预测鱼露发酵过程中TVB-N及AN质量浓度。

The changes of physical and chemical indicators were studied during the fermentation of fish sauce.The contents of volatile base nitrogen(TVB-N)and amino acid nitrogen(AN)were predicted by analyzing the indicators.130 sets of experimental data were obtained by testing the physical and chemical indicators under different fish sauce fermentation conditions.Through the correlation analysis of Person,TVB-N and AN contents were positively correlated with fermentation time and salt concentration,negatively correlated with pH,and didn’t relate to temperature and the amount of koji.The PCR,PLS and BP neural network models were applied to the prediction of TVB-N and AN contents.The results showed that the three models could predict TVB-N and AN contents of fish sauce fermentation process.Comparing the prediction parameters of the three models,it was found that the BP model had lower RMSEP and higher R2 and RPD.The results showed that the BP model could predict TVB-N and AN contents more accurately during the fish sauce fermentation process.