Study And Application Of Smartnose On Flavor Quality Identification Of Tartary Buckwheat Tea

Flavor is the main factor in evaluating the quality of TBT. The flavor characteristic ofTBT is affected by a number of aroma components rather than only one simple substance.Electronic Nose is a new hi-technology, which imitates human’s olfaction and detects thevolatile components based on selective sensor array. The fingerprints got after detection couldrepresent the entire information of sample’s volatile components. Many studies reported theapplication of Electronic Nose on tea quality evaluation, and it has a great potential inevaluating flavor characteristics of TBT. Based on Electronic Nose technology, the paper usesa variety of pattern recognition methods to study the application outlook of Electronic Nosetechnology on discriminating different Tartary Buckwheat Tea’s flavor quality.The main results are as follows:(1) Results of Loading showed that sensory response values of Electronic Nose weredifferent among various TBT samples which indicated the differences existed in flavorcomponents of different samples. Also, the Principal Component Analysis (PCA) of dry teaand tea infusion sample in same group showed obvious differences, which indicated that afterpouring boiling water on TBT, the flavor characteristics, aroma composition and quantitychanged significantly, which could directly affect the result in pattern recognition.(2) Results of pattern recognition of TBT samples made by different process (includingthree by whole grain roasting process, six by puffing process and six by broken rice roastingprocess) showed that the difference in flavor characteristics among tea infusion samples wasmore obvious than dry tea samples, therefore the results of Electronic Nose response value inpattern recognition are more accurate and efficient. Also, in the results of PCA, thecumulative contribution rate of first three principal components reached more than85%andthey were able to represent original information. However, discrimination of samples by PCAwas unsatisfied, while the discrimination rates in Fisher Discriminatory Analysis (FDA) andMultilayer Perceptron Network (MLPN) were much higher. Furthermore, all samples tested,except only two products using puffing process, the results of which were overlapping andcannot be discriminated efficiently by FDA and MLPN, could be discriminated entirely.The results indicated that with the effects of different processing technology, parameterand raw material, TBT samples of various brands presented different flavor quality.Electronic Nose technology could be applied to identify flavor components and patternrecognition method of FDA and MLPN could effectively discriminate, which is of greatimportance in discerning the false from the genuine in TBT market and helping to developndustry standard. (3) Pattern recognition results of TBT in different roasting process showed thatElectronic Nose could not clearly discriminate three kinds of dry tea roasted below160℃aswell as PCA, FDA, MLPN. However, tea infusion samples could be accurately discriminatedby FDA and MLPN. While flavor characteristics of both dry tea and tea infusion of threesamples roasted above160℃were significantly different and could be discriminated byFDA and MLPN. The above results showed that firstly, when roasting below160℃,temperature had few influence on flavor components in dry TBT, while when the roastingabove160℃, constituents of aroma components changed with the temperature. Secondly,Electronic Nose response values of tea infusion samples which roasted in15,20,25min at160,170,180℃respectively could be discriminated entirely through FDA and MLPN. As todry tea, only the three samples processed in180℃could be discriminated by FDA andMLPN, which indicated that the influence of roasting time on flavor characteristic was greaterunder higher temperature. By contrast, flavor characteristic of tea infusion sample was moreeasily to be identified by Electronic Nose.In summary, the flavor quality of TBT made by whole grain roasting process wasobviously influenced by roasting temperature and time, and Electronic Nose technology andpattern recognition could be used to discriminate among different samples. Therefore, strictcontrol of roasting parameters and normalized processing standard are essential to ensurestable quality of TBT.