HIDUNG ELEKTRONIK BERBASIS SENSOR TGS UNTUK PENGELOMPOKKAN KESEGARAN IKAN AIR TAWAR; TGS BASED ELECTRONIC NOSE FOR CLASSIFICATION OF FRESH WATER FISH FRESHNESS

ASHARUDIN ACHZAB, Triyogatama Wahyu Widodo

2015 | Skripsi | FMIPA

A research on the grouping freshness of freshwater fish using an electronic nose is conducted. This system uses TGS-based gas sensor array to recognize the aroma of fresh-water fish freshness. Grouping freshwater fish freshness is very important because of the freshness of the fish is a determinant of the price of fish in the market. Currently, human nose is used as a tester to predict the fish freshness. However, in the reality this method still has some weaknesses such subjectivity, while the use of standard chemical analytical methods are expensive and requires experts to represent them. The needs of the electronic nose that can detect the freshness of the fish based aroma is indispensable. Data matrix samples obtained were then classified using PCA and freshwater fish freshness prediction using neural networks Back Propagation. Classification of the PCA produces 99.8% variance for pomfret and catfish in the freezing temperatures, tilapia for freezing temperatures have a 99.5% variance. Pomfret at room temperature produces 99.7% variance value, catfish at room temperature produces variance value 99.6% and room temperature tilapia produces 99.9% variance. While predictions with ANN-BP showed the results is close to the results predicted that it should with the proper network configuration.

Kata Kunci : neural networks; PCA; fish aroma