<![CDATA[This research aims to develop a method for identifying gas types based on multisensor data using Recurrent Neural Network (RNN) in the context of Electronic Nose (E-Nose) application. The method utilizes Quartz Crystal Microbalance (QCM) sensors that respond to changes in oscillation frequency to detect gases. The data used in this study were obtained from QCM sensor measurements on six species of mint at the Botanical Institute of Karlsruhe Institute of Technology (KIT), Germany, recorded by Shalih Okur. Through the training process using RNN models with ReLU and LeakyReLU activation functions, training accuracies of 98.84% with a computation time of 326 seconds (ReLU) and 97.78% with a computation time of 267 seconds (LeakyReLU) were achieved. Furthermore, in the identification phase, the RNN model achieved accuracies of 79% with a computation time of 10 seconds (ReLU) and 85% with a computation time of 4 seconds (LeakyReLU). These findings indicate the potential of the RNN method for gas type identification based on multisensor data, with a focus on QCM sensor usage. Thus, the results of this study demonstrate the effectiveness of the RNN method in identifying gas types based on multisensor data, particularly when utilizing QCM sensors. Keywords: Multisensor, Recurrent Neural Network (RNN), Gas identification]]>
Copyrights © 2023
