Aplikasi Electronic Nose Berbasis Sensor MQ-2 dan MQ-9 dalam Deteksi Adulterasi Daging Babi pada Produk Bakso Sapi

Abstract

The adulteration of beef meatballs with pork is a serious issue in Indonesia. As the majority of the population is Muslim, compliance with halal food standards is a fundamental requirement that must be guaranteed. This practice is driven by economic reasons and the desire to enhance flavor, yet it directly impacts the rights and health of Muslim consumers. In this study, an electronic nose (E-Nose) system was developed based on MQ-2 and MQ-9 sensors and optimized using supervised learning, such as Support Vector Machine (SVM), Convolutional Neural Networks (CNN), and Bidirectional Long Short-Term Memory Convolutional Neural Networks (BiLSTM-CNN), as a rapid and non-destructive testing method to detect pork adulteration in beef meatballs. Seven variants of beef meatball samples with pork content of 0% (negative control), 7%, 15%, 22%, 30%, 37,5%, and 75% (positive control) were evaluated in a one-factor experimental design with three replicates. Detection was performed at 70 °C using an E-nose via a flushing-sensing-cleaning procedure, and the output data was processed through interloop, normalization, statistical feature extraction, and classification using Python 3-based Google Colab. The MQ-2 and MQ-9 digital signal patterns exhibit a consistent response curve during sensing phase, with the gas concentration (ppm) rising sharply, indicating optimal sensor performance. All three models achieved 100% accuracy for samples A1, A2, A3, A5, and A6. However, for sample A4 (15% pork), the accuracy of SVM and CNN dropped to 99.5%, whereas BiLSTM-CNN consistently maintained 100% accuracy. These results demonstrate that the e-nose, based on MQ-2 and MQ-9 sensors optimized by supervised learning, is a device capable of quickly and accurately detecting adulteration even at low levels (7%).