Identification of the Origin of Zanthoxylum Bungeanum by Electronic Nose Combined with Fuzzy Neural Network

摘要

The origin of Zanthoxylum bungeanum was quantitatively analyzed and identified by electronic nose combined with optimized neural network. Seven different electronic noses were used to collect information about four kinds of Zanthoxylum bungeanum odors. The effective variables were extracted by one-way Analysis of Variance (ANOVA). The neural network algorithm was used for pattern recognition, and the fuzzy control rules were used to optimize the neural network. The results show that one-way ANOVA has a good effect in selecting effective variables. BP neural network based on the selected variables can effectively distinguish the origin of Zanthoxylum bungeanum. The recognition rate of test set is 92.14% as well as the verification set is 90%. The recognition rate of test set of BP neural network optimized by fuzzy control is 97.86%, and the recognition rate of verification set is 100%. The optimized classification results show that the fuzzy-BP neural network can better identify the origin of different varieties of Zanthoxylum bungeanum. It is indicated that the electronic nose combined with fuzzy neural network has certain feasibility for identifying the Zanthoxylum bungeanum producing area.