| Summary: | In Indonesia, there are two types of coffee grown i.e Robusta and Arabica.
Arabica coffee has higher quality flavor than Robusta, but it has less production. If
both types are combined, there fore probably produce a good flavor quality with high
taste and intense color. The distinctive aroma come from roasting process, which lead
to physical and chemical changes in the coffee beans. Usually, organoleptic
evaluation of coffee roasted is traditionally done on human senses. After roasted, the
aroma quality of coffee grown is decreased quickly, so it is necessary to have good
storage conditions and proper packaging. Evaluating the aroma quality of coffee
grown use Electronic Nose that has four sensor i.e TGS 822, TGS 825, TGS 826 and
TGS 2602. Today, Artificial Neural Network (ANN) is an attractive tool for
modeling complex processes. This study aims to classify Arabica and Robusta coffee
blends that were roasted and then stored in different packaging variations based on
the results of the quantitative value Electronic Nose using pattern recognition ANN
and PCA. This research used 3 treatment i.e blending, package and temperature
storage. The percentage Arabica : Robusta blends coffee were used respectively:
100:0%, 75:25%, 50:50%, 25:75, 0:100%. 3 types of packages were used i.e :
aluminum foil, paper and glass. 2 temperature storages are cool temperatures (15 °C)
and absolute temperatures (27°C). ANN used bacpropagation method, the MSE 1 x
10
-1
, maximum epoch 300000, 4 pieces hide layer. 70% of the data were used for
training and 30% for testing. The maximum blends Arabica and Robusta is 75 % :
25% with grade 1. Aluminum foil packaging are better in protecting the coffee
powder during storage compared to paper and glass packaging. Grown coffee is better
stored in cool temperatures compared to absolute temperature. ANN is able cluster
blends coffee Arabica and Robusta coffee with R value = 0.97 and the accuracy value
75% for recognizing the type of coffee blending and 70.6% for recognizing of coffee
during storage. PCA can reduce the data so there is no redundancy in the Arabica and
Robusta coffee blending with the aroma sensor input value.
|
|---|