Pengembangan Klasifikasi dan Identifikasi Jenis Daging Menggunakan Deret Sensor dan Jaring Saraf Tiruan (JST)

Abstract

Chicken, beef, and lamb are classified based on the distinctive aroma of the meat determined using an electronic nose consisting of a series of sensors with six pieces. This electronic nose also comprises a Conducting Polymer, which has a high resistance and is widely used as an insulator. However, this resistance has a specific limit where the polymer surface turns into carbon and conducts electric current when exposed to excessive electric charge. This research was conducted by placing chicken, beef, and lamb samples in a closed container at room temperature. Data were collected on the first, second, and third days alternately from the odor of chicken, beef, and lamb. A Neural Network (NN) for pattern recognition and an ATMega16 microcontroller for data acquisition were further conducted. NN was trained using Kohonen and a Conducting Polymer sensor with a voltage used as the output to generate changes in the polymer resistance. A two-layer NN consisting of six input nodes and three output neurons was also trained using the Kohonen algorithm, which was completed in 31 iterations. The test was carried out 30 times for each exposure of chicken, beef, and lamb with steam, and a successful system percentage of 93.33% was obtained.