Classification of different quality black tea in the same and separate harvest period using e-nose

An example of the most consumed beverage in the world is tea and our country has an important place in terms of the tea sector. Therefore, the quality of tea is always a priority issue. It is desired that tea products are produced with healthy methods and have a good quality standard. Before the tea is placed on the market for sale, it must demonstrate accuracy in the intended quality. The subjective decision of people in determining the standards of tea quality can lead to variable results. Controlling the quality of tea with objective methods is necessary to achieve more accurate results in tea factories. In this thesis, a minimally designed electronic nose prototype was used for the classification of the quality of black teas with different qualities in the same and separate harvest periods produced in the Eastern Black Sea Region, based on the scent of tea. In this e-nose circuit, a systematic way was followed and the smell of black tea was detected by 8 different sensors. The voltage change caused by the sensors operating based on the conductivity change was recorded on the data acquisition card and displayed digitally on the computer. These signals obtained with the e-nose circuit were preprocessed, feature extraction was performed and k-nearest neighbor, support vector machines and bagging learning algorithms were used for classification. The 5-fold cross-validation method was used to examine the classification achievements and the percentages of success in the results of the method were determined and compared.