IDENTIFIKASI BAHAN BAKU ETANOL DENGAN HIDUNG ELEKTRONIK MENGGUNAKAN METODE PRINCIPLE COMPONENT ANALYSIS DAN JARINGAN SARAF TIRUAN BACKPROPAGATION; RAW MATERIAL ETHANOL IDENTIFICATION USING ELECTRONIC NOSE WITH PRINCIPLE COMPONENT ANALYSIS AND ARTIFICIAL NEURAL NETWORK-BACKPROPAGATION

Hamidi, Febrian, Danang Lelono

2015 | Skripsi | FMIPA

Various contributions of ethanol in human life make ethanol to be one of the required materials. Ethanol can be used as an antiseptic and an alternative solution / mixture of fossil fuel. One way to get ethanol is from fermented plant containing starch, sugary and fibrous like cassava, cassava, sugar cane and corn. Each of these materials produce ethanol qualities are different. Quality testing for ethanol using standard chemical analytical GC (Gas Cromatograph). This system has a high accuracy, but the costs, time and experienced experts in operating it. That requires instruments that can identify raw materials ethanol based flavor/aroma. E-Nose is an instrument of accession which can detect the scent, which the system uses a sensor array that changes the flavour gas into the response signals. The response signal is converted to forms of digital and extracted characteristics with maximum signal response method. Maximum signal response is then grouping using the PCA method and identified using ANN-BP. Flavour is ethanol samples of each ingredient maize, cassava, cassava and sugarcane as much as 20 ml is placed into the sample chamber. Then do the inhalation process (odor on) by passing the aroma of the sample to the surface of the sensor array and disposal process (odor off) in the form of cleaning the sensor surface of the flavour. The process of inhaling is done repeatedly. The test results obtained by grouping data using PCA produces two main components of the percentage variation of 98.3%. While the identification using ANN-BP obtained test accuracy rate of 94.2% and MSE of 0.718667 using the optimum configuration. With so it can be said that this research has identified variation with E-nose ethanol using PCA and ANN-BP.

Kata Kunci : Artificial Neural Network (ANN); Principle Component Analysis (PCA); Backpropagation; electronic nose; ethanol