Abstract: A rapid and accurate method for the classification of Mee tea grades based on olfactory visualization technology was proposed. First, 12 different porphyrin indicators which were found to have an obvious chromogenic effect in the preliminary experiment were selected to develop a color-sensitive gas sensor array. The sensor array was then used to test different grades of Mee tea infusion, and the characteristic images were acquired and analyzed by principal component analysis (PCA) for dimensionality reduction. The obtained PCA results with different dimensionalities were used as input variables to establish a classification model for Mee tea grades using support vector machine (SVM). Finally, three swarm intelligence algorithms, firefly algorithm (FA), gray wolf optimization (GWO) and cuckoo search (CS), were used to optimize the penalty factor (c) and the kernel function parameter (g) of the SVM model. Results showed that the classification accuracy of the original SVM model for the test set was 80% with 12 principal components needed, and increased after optimization with each of the three algorithms. The classification accuracy of the model optimized by CS was 93.3%, and the number of principal components needed was reduced to 6. Accordingly, olfactory visualization technology can be used to classify Mee tea grades, and the model performance can be enhanced and the classification accuracy can be improved by swarm intelligence optimization algorithms significantly.

Key words: Mee tea; olfactory visualization; grade classification; support vector machine; swarm intelligence optimization algorithm