Main Article Content
Abstract
Abstract
The research aimed to design the systems of tropical fruit maturity identification based on electronic nose using Array SnO2 semiconductor gas sensor. The research utilized five TGS sensors, namely TGS2600, TGS2602, TGS813, TGS2611, and TGS2612. The array sensor outputs are acquired by personal computer through interface unit based on microcontroller Atmega 8535. The acquisitions are made every 0.5 seconds for a minute for each sensor output. Then, it was determined the average sensor output as an input for Artificial Neural Network (ANN) which used Multi Layer Perceptron (MLP) architecture with three layers. ANN Training applied Backpropagation algorithm. The results showed the sensor output responses vary by the level of maturity of fruit. The obtained training yielded the architecture of ANN for the fruit maturity identification system were 5 inputs and 4 outputs with a number of hidden layer neurons for oranges and strawberries was 16 while for tomatoes was 32. The identification application showed that the successful identification percentage of orange was 93.75%, 75% of strawberries, and 81.25% of tomatoes. Overall success rate of detecting the level of maturity of fruit (oranges, strawberries, and tomatoes) was 83.33%.
Keywords: E-nose system, TGS sensor, fruit maturity, ANN application
Abstrak
Penelitian bertujuan merancang bangun sistem identifikasi kematangan buah tropika berbasis penciuman elektronik (e-nose) menggunakan deret sensor gas semikonduktor SnO2 menggunakan jaringan syaraf tiruan. Dalam penelitian digunakan deret 5 sensor seri TGS: TGS2600, TGS2602, TGS813, TGS2611, dan TGS2612. Deret sensor diakuisisi dalam komputer melalui unit antarmuka berbasis mukrokontroler Atmega 8535. Akuisisi dilakukan dalam 1 menit tiap 0.5 detik sehingga diperoleh 120 data untuk tiap keluaran sensor. Ditentukan rata-rata keluaran sensor sebagai masukan Jaringan Syaraf Tiruan (JST). Arsitektur JST menggunakan Multi Layer Perceptron (MLP) dengan 3 lapis. Hasil penelitian menunjukkan respon keluaran sensor berbeda-beda untuk tiap tingkat kematangan buah. Pelatihan JST menggunakan algoritma backpropagation. Dari hasil pelatihan didapatkan srsitektur jaringan syaraf tiruan untuk sistem identifikasi adalah 5 input dan 4 output dengan jumlah neuron hidden layer untuk identifikasi kematangan jeruk dan stroberi adalah 16 sedangkan untuk tomat adalah 32. Dari hasil pengujian aplikasi diperoleh persentase keberhasilan identifikasi kematangan buah jeruk sebesar 93.75%, stroberi sebesar 75%, dan tomat 81.25%. Secara keseluruhan persentase keberhasilan sistem dalam mendeteksi tingkat kematangan buah (jeruk, stroberi, dan tomat) adalah sebesar 83.33%.
Keyword: sistem penciuman elektronik, sensor TGS, kematangan buah, aplikasi JST
Diterima: 04 Oktober 2010; Disetujui: 28 Februari 2011
Article Details
Authors submitting manuscripts should understand and agree that copyright of manuscripts of the article shall be assigned/transferred to Jurnal Keteknikan Pertanian. This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA) where Authors and Readers can copy and redistribute the material in any medium or format, as well as remix, transform, and build upon the material for any purpose, but they must give appropriate credit (cite to the article or content), provide a link to the license, and indicate if changes were made. If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.