Design And Development Of A Microcontroller-Based Air Pollution Monitoring System At Traffic Light Areas Using The Mamdani Fuzzy Logic Method

Abstract

Air pollution was a serious issue in Indonesia, particularly in areas with high vehicle density, such as Yogyakarta. In 2024, more than 164 million vehicles were recorded in Indonesia. Motor vehicle emissions contributed to approximately 85% of total air pollution in the country. The high activity of motor vehicles increased the concentration of harmful gases such as carbon monoxide (CO), which had adverse effects on public health. To assess air pollution levels in surrounding areas, a monitoring system of air quality was required to maintain acceptable environmental conditions. In this study, a system was designed by integrating the MQ-135 sensor to detect carbon monoxide (CO) in real-time. The data obtained from the sensor were transmitted to the ESP32, where data processing employed the Mamdani Fuzzy Logic method to classify air quality into three categories: Good, Moderate, and Poor. The process began with CO level detection and was followed by fuzzification, inference, and defuzzification stages, producing a crisp value as a reference for air quality classification. In addition, the system was equipped with a monitoring feature based on Firebase and an application developed with MIT App Inventor. Based on the results of testing, the air pollution monitoring system was successfully implemented. The MQ-135 sensor was able to read surrounding air conditions in real-time with an average error of 0.33%, resulting in a reading accuracy level of 99.67%. The Mamdani Fuzzy Logic method effectively processed ambiguous data; however, testing also identified limitations in the design of membership functions, which led to mismatches in category determination at threshold values. The system successfully transmitted data in real-time to Firebase and the MIT App Inventor application with a success rate of 80%.