The establishment of ham grade, sensory scores and key flavor substances prediction models for Jinhua ham via E-nose combined with machine learning

Abstract

Unique organoleptic and flavor attributes of Jinhua ham are associated with their qualities. However, methods for quickly predicting the grade of hams, sensory scores and key flavor substances have not been systematically established. This study used sensory evaluation and E-nose to analyze the sensory differences for different grades of Jinhua ham. GC–MS was combined with O2PLS and correlation analysis to identify the key flavor substances. Classification model based on E-nose response signals was established by logical regression to predict the ham grades, which displayed a high classification performance, with the accuracy of 0.87. Moreover, linear regression and random forest models were established to predict the sensory score of hams and the concentrations of key flavor substances, meanwhile the R² were all above 0.7, demonstrating the model applicability. This work provides a theoretical basis for rapidly predicting qualitative and quantitative parameters of Jinhua ham by E-nose with machine learning.