Development of smart air pollution monitoring with machine learning forecasting system for smog in Quezon City. 6

Undergraduate Thesis : (Bachelor of Science in Computer Science) - Pamantasan ng Lungsod ng Maynila, 2024. 56

5

STATEMENT OF THE PROBLEM: 1. Outdated or Unavailable Pollution Monitoring : According to DENR, Quezon City currently lacks a modern, efficient system for monitoring air pollution, which hinders the ability to accurately assess the environmental and health impacts. 2. Lack of Accessible Data: Due to inadequate monitoring systems, there is a significant gap in reliable data regarding air quality in Quezon City, making it challenging to understand the full extent of the pollution problem. 3. Limited Proactive Measures by Local Government Units (LGUs): The absence of comprehensive data and effective monitoring tools impedes the ability of LGUs to formulate and implement proactive strategies to combat air pollution and protect public health. RESEARCH METHODOLOGY: This research employed a mixed-methods approach, utilizing both qualitative nad quantitative data. A literature review informed the project design, while procedures encompassed: . Website development: Utilizing React JS for the user interface and Python with Flask for the backend. This website seves as a public platform for education and data visualization. . AI model development: Leveraging historical data from PAG-ASA and DENR to train a Long Short-Term Memory (LSTM) model for among prediction. . Hardware design and integration: Constructing an air quality monitoring system with an Arduino Nano ESP32 microcontroller and various sensor modules (DHT11, PMS7003, SGP40, MQ131, MQ7) to collect real-time environmental data. These procedures ensured the creation of a comprehensive air quality monitoring system, yielding valuable data that contributes to achieving the project's goal. The system, deployed in La Loma, Quezon City, Philippines, empowers residents with real-time air quality information and smog predictions. SUMMARY OF FINDINGS: This study successfully developed a smart air quality monitoring system for La Loma, Quezon City. The system utilizes various sensors to collect real-time data on air quality parameters like temperature, humidity, particulate matter, ozone, carbon monoxide, and VOCs (volatile organic compounds). This data is collected every hour for a 13-hour period. The system goes beyond data collection by incorporating machine learning for smog prediction. This allows residents to received advanced warnings about potential among events, empowering them to take necessary precautions, particularly those with respiratory issues. The accuracy of the sensor readings was verified through companions with commercially available devices, demonstrating a high degree of reliability. Additionally, a user-friendly platform allows residents to easily access this real-time air quality information, enabling them to make informed decisions about their health and activities. CONCLUSION: In conclusion, the researchers successfully built a smart air pollution monitoring system with real-time data, machine learning-based among prediction, and a user-friendly interface. This system empowers residents with air quality information and forecasts, potentially leading to improved public health and cleaner air in Quezon City. Limitations like accuracy and wider adoption need to be addressed for future improvements. RECOMMENDATIONS: While resource limitations (time and budget) were encountered, the project paves the way for future advancements. To enhance the system's impact, future research should focus on several key areas. Firstly, translating air quality data into real-time health risk assessments for different demographics would empower residents to take targeted protective measures. Secondly, expanding the monitoring system into a city-wide or even regional network would provide a more comprehensive understanding of air pollution patterns and potential cross-city effects. Thirdly, exploring deeper learning architectures like RNNs and CNNs for smog predition holds promise for even more accurate forecasts. In addition, the future researcher may also use sim card module for hardware circuit for remote areas as well as a threshold just in case to protect the device from unnecessary disturbances and a solar panel as a power source. Finally, researchers may add a features like alert system for incoming smog and rising of temperature and a notification when the ppm risks into hazardous environment and make it a mobile app for case of convenience.

5