Because sensory analysis and chromatographic analysis were not well suitable for the discrimination between Chinese Jing wines and counterfeits, an electronic nose (in short, eNose) was employed to carry out the task. In the investigation three kinds of features of eNose signals were extracted and as input data of principal component analysis (PCA). These features are named as mean-differential coefficient value (MDCV), energy value of wavelet packet decomposition (WE) and relative steady-state response value (RSV), respectively. The results demonstrated that the discrimination based on these features data could all be performed by PCA, and the RSV was the best. At the same time, an evaluation method was proposed to evaluate the discrimination capability of these features quantitatively, and the evaluation results are basically in accord with PCA discrimination results. This showed the evaluation method was appropriate for evaluating the discrimination capability of different features. In conclusion, the investigation indicated that the eNose coupled with PCA was absolutely competent for the discrimination tasks, and especially the feature RSV was simple and reliable.
展开▼
