Deep Neural Network Method to Classify Empon-Empon Herb Based on E-Nose

Indonesian herbal drink, called Jamu, is one of the cultural heritage herbal drinks which is one of the characteristics of Indonesian culture. The raw material for herbal medicine is the herbal plant that has many benefits with distinctive characteristics, i.e. color, smell, and texture. In Indonesia, there are types of raw materials for herbal medicine, called empon-empon, galangal, and turmeric which are similar in color, shape, and smell. Therefore, ordinary people sometimes difficult to classify. In this study, the types of empon-empon based on their smell were classified into four classes, namely, ginger, galangal, and turmeric-based on their odor. The smell of the empon-empon is obtained from the e-nose which designed using the TGS2611, TGS813, and MQ136 sensors connected to the Arduino Uno. The smell characteristic of empon-empon is used as the value of the sensor voltage. The voltage values that have been obtained are classified using a deep neural network. Based on the results of the classification, it is found that the deep neural network can classify the types of empon-empon based on odor with an accuracy of 86%.