IDENTIFIKASI CAIRAN MUDAH TERBAKAR DAN CAIRAN TIDAK MUDAH TERBAKAR DENGAN ELECTRONIC NOSE MENGGUNAKAN METODE PRINCIPAL COMPONENT ANALYSIS DAN JARINGAN SYARAF TIRUAN BACKPROPAGATION; IDENTIFICATION OF FLAMMABLE LIQUIDS AND INCOMBUSTIBLE LIQUIDS WITH ELECTRONIC NOSE USING PRINCIPAL COMPONENT ANALYSIS METHOD AND NEURAL NETWORK-BASED USING BACKPROPAGATION

Astriyanie, Risha Putrie, Danang Lelono

2015 | Skripsi | FMIPA UGM

During this time to detect flammable liquids are still using selective gas indicator. However, these methods require a multitude of sensors selectively involved along with increasing types of liquids, besides that not all hazardous gases can be detected due to the limitations of the selective gas sensor. For that, it require an instrument that be able to detect aroma of flammable liquids and incombustible liquids using gas sensor array unselected combined with the pattern recognition. The instrument known as electronic nose that works like a human nose in the inhalation and exhalation. Inhaled (collecting) and exhaled (purging) aroma samples respectively carried for 120 seconds with 10 peaks data collecting a day in 10 days. Stimulus aroma captured by unselected sensor array into signal overlapping in the form of digital fingerprint then extracted using differential method and claasified with PCA before identified by ANN-BP. Classification by PCA generate a cumulative percentage with variation of two major components of 93,3%. While the identification of samples with the ANN-BP obtained 99,3% accuracy rate with the MSE (Mean Square Error) 0,0066729 on all 192 epochs. It was proved that flammable liquids and incombustible liquids has been successfully performed using e-nose (electronic nose) with PCA (Principal Component Analysis) method and backpropagation neural network (ANN-BP).

Kata Kunci : Electronic nose; Principal Component Analysis; ANN-BP; flammable liquids; incombustible liquids