Faster predicting the content of key non-volatile compound in rosemary using electronic nose with multivariate algorithms

To establish a rapid and simple method for predicting the content of key non-volatile compounds in rosemary, compounds from rosemary were analyzed using an electronic nose (E-nose) with 18 sensors (S1-S18), headspace solid phase microextraction combined with gas chromatography-mass spectrometry (HS-SPME-GC-MS), and liquid chromatography-mass spectrometry (LC-MS). The data were analyzed by cluster analysis, principal component analysis (PCA). A total of 161 volatile compounds were detected using GC-MS, including 40 alcohols, 2 aromatic hydrocarbons, 5 phenolic compounds, 2 furan compounds, 1 sulfur compound, 6 ethers, 6 aldehydes, 2 acids, 5 terpene, 19 ketones, 16 esters, and 57 other compounds. The content of caffeic acid, nepetin, luteolin, apigenin, diosmetin, rosmarinic acid, carnosic acid, and rosmanol in rosemary samples was determined using LC-MS. The odor profile of rosemary was analyzed using the E-nose. The PCA indicated using the E-nose for discriminating the quality of rosemary was feasible. Meanwhile, the partial least squares (PLS) and artificial neural networks (ANN) model for predicting the content of key non-volatile compounds in rosemary was established using E-nose. In comparison with the PLS model, the constructed ANN model possessed greater predictive capability. Predicting the content of non-odors from rosemary using odor detector was feasible. This provides a basis for the rapid detection method for rosemary using an E-nose.