Study On Rapid Brand Recognition Of Apple Essence Based On Multiple Classifiers

Efficient rapid essence detection method to ensure safety of flavoring essence on the market is of great significance.Fingerprint technology combined with pattern recognition method for the quality assessment of essence is widely used.Effective spectrogram screening and identification algorithm building are key tasks of pattern recognition.Analysis on essence quality stability and identification of different essence brands are fundenmental purposes.Similarity analysis method is able to filter outliners spectrograms,and analyse essence quality stability of different manufacturers and different batches.Artificial neural network?K nearest neighbor?decision tree are three excellent machine learning algorithm.In this paper,apple essence brands classification is studied on the basis of similarity analysis and neural network?K nearest?decision tree.Ion Mobility Spectrometry Technology and pattern recognition technology are combined to establish a model for quality evaluation of apple essence.At first,we collected Ion Mobility Spectrometry fingerprint chromatography of five kind of apple essence samples provided by three famous domestic flavoring essence companies,to be directed against the outliners came from volatility of ion mobility spectrometry equipment test,we used similarity analysis method to discriminate outliners.Then we identified different kinds of apple essences using artificial neural network?k-nearest neighbor algorithm?decision tree and achieve good classification effect.Recognition rates were as high as 99.41%?98.82%?96.47%?To further simulate quality analysis of apple essence from different manufacturers and different batches in actual production,we collected Ion Mobility Spectrometry fingerprint chromatography of four kind of apple essence samples provided by four famous domestic flavoring essence companies and integrated them using data fusion technology.Then we carried out similarity analysis and identification of positivenegative?integrated mode apple essences.Experimental results show that essences of different batches have quality fluctuation to some extent.In data fusion mode,can reflect more comprehensive information of essences and improve the recognition rate of essences of different manufacturers.Recognition rates were as high as 99.17%?100%?96.42%?In this paper,Ion Mobility Spectrometry method is used combined with similarity analysis method?neural network?K nearest neighbor and decision tree classification method to establish rapid and accurate spirit brand identification method,designed to provide effective technical support for apple essence quality detection.