Abstract:The odor of citrus juice changes during the storing process. We use an electronic nose (E-nose) to detect the volatile odors released by citrus juice and use the detect results to classify citrus juices from different storage periods. In this article, a novel classifier of E-nose, namely broad learning system (BLS) is introduced. BLS is different from traditional classifier. It has a simple network model, which can greatly reduce the training time of the model. BLS can effectively combine feature extraction and classification recognition to make the model more efficient. We apply BLS to the analysis of valencia citrus juice data. The experimental results show that BLS can effectively identify the current stage of the stored valencia citrus juice. Compared with traditional classifier such as radical basis function neural network (RBFNN) and linear discriminant analysis (LDA), the results show that BLS has better performance for the storage period classification of valencia citrus juice.

A Classification for Electronic Nose Based on Broad Learning System