Development of a portable electronic nose based on a hybrid filter-wrapper method for identifying the Chinese dry-cured ham of different grades

In this work, a portable electronic nose (e-nose) equipped with a smartphone was developed to identify the Chinese dry-cured ham of three grades. The gas chromatography-ion mobility spectrometry (GC-IMS) was employed for detection of the volatile organic compounds of hams and optimization of the sensor array. A hybrid filter-wrapper method was proposed to optimize the feature set which included the time and frequency domain features. The proposed hybrid method included two parts: the filter method based on mutual information mixed evaluation (MIME) which was applied to eliminate the irrelevant features, and the wrapper method based on support vector machine-backward feature elimination with cross-validation (SVM-BFECV) which was applied to removing the multicollinear features. Both the principal component analysis and T-distribution stochastic neighbor embedding with the hybrid filter-wrapper method presented good results, and all the samples could be classified completely. SVM, K-nearest neighbors and logistic regression were applied for the prediction works. SVM based on the hybrid method presented the best results, and the prediction accuracy and consuming time was 96.06% and 17.32 s, respectively. Above all, the proposed filter-wrapper method performed well in optimizing the feature data, and the three grades of hams can be clearly identified by using the developed portable e-nose based on the optimized features.

Automated summary

In this study, a portable e-nose equipped with a smartphone was developed to identify different grades of Chinese dry-cured ham using GC-IMS technology and a hybrid filter-wrapper method for feature optimization. SVM algorithm yielded the best prediction results, with an accuracy of 96.06% and a time consumption of 17.32 s. The filter-wrapper method performed well in optimizing feature data and enabling clear identification of ham grades.