Development of wavelet transforms to predict methane in chili using the electronic nose

Chili (Capsicum annum L.) became one of the most popular fruits. In Indonesia, there are various types of chili, but the most commonly purchased by the community are cayenne pepper (Capsicum frutescens) and green cayenne pepper. One of the gases in chili is methane. When the body is exposed to methane, the oxygen level will decrease and the lungs will lack oxygen that causes dehydration, resulting in a serious impact on the body. This study aims to detect the type of chili that has the most methane content by using E-Nose. One of the E-Nose is an MQ-6 sensor. This sensor can detect, analyze and distinguish methane content. There are seven stages of this research that are, proper sensor selection, training data retrieval that will be used as groundtruth data, signal readout from sensor, noise reduction using low pass filtering and wavelet transform, then normalized with Z-Score, compare Support Vector Machine with Linear Discriminant Analysis, and calculate accuracy, precision, sensitivity with confusion matrix 20 times iteration. This Study obtained an accuracy of 93% with two classes of cayenne pepper and green cayenne pepper using Linear Discriminant Analysis and 92% with Support Vector Machine. To convert the signal frequency to ppm using Artificial Neural Network obtained ppm. The result is value from green cayenne pepper of 2000 to 6000 ppm whereas in cayenne pepper under 2000 ppm. From the results of experiment's chili that contains the most methane is green cayenne pepper.