Real-Time Air Quality Prediction Using Metrologically Calibrated Gas Sensors and Random Forest Algorithm

Abstract

The increasing level of urban air pollution requires monitoring system that are capable not only of measurement but also real time prediction. Low coast gas sensor such as MQ-135 are widely used due to their affordability and ease of integration. However, these sensors exhibit limitations in terms of accuracy, signal stability, and drift characteristics. This research proposes a real time air quality prediction model based on gas sensor data using a machine learning approach integrated with metrological calibration. The system consists of a microcontroller base data acquisition module, aserver for data storage, and a predictive model deployed for real time computation. Data were collected over a controlled observation period with fixed sampling intervals. Preprocessing steps included regression based calibration, min max normalization, and noise reduction using a movig avarage filter. Three algorithms were evaluated Linear Regression, Random Forest, and Long Short-Term Memory. Model performance was assessed using Root Mean Square Error, Mean Absolute Error, and coefficient of determination. The results indicate that the Random Forest model achieved the lowest RMSE and demonstrated stable prediction performance under sensor signal fluctuations. The integration of calibration prior to model training significantly improved prediction accuracy compared to models without metrological correction. The proposed system provides reliable real-time air quality prediction and can support intelligent environmental monitoring and local decision-making processes.

REFERENCES

Alahi, M. E. E., Sukkuea, A., Tina, F. W., & Mukhopadhyay, S. C. (2020). Integration of IoT-enabled technologies for air quality monitoring and prediction. IEEE Internet of Things Journal, 7(10), 9871–9882. https://doi.org/10.1109/JIOT.2020.2994523

Chen, J., Li, X., Wang, Y., & Zhang, H. (2022). Comparative evaluation of machine learning models for air pollution forecasting. Atmospheric Environment, 268, 118804. https://doi.org/10.1016/j.atmosenv.2021.118804

Esposito, E., De Vito, S., Salvato, M., & Bright, V. (2021). Dynamic calibration of low-cost air quality sensors using machine learning techniques. Sensors, 21(12), 3989. https://doi.org/10.3390/s21123989

Gao, L., Zhang, D., & Li, J. (2020). Calibration and drift compensation of gas sensors using data-driven models. Sensors and Actuators B: Chemical, 305, 127451. https://doi.org/10.1016/j.snb.2019.127451

Hernandez, W., & Garcia, R. (2021). Data preprocessing strategies for improving air quality prediction accuracy. Environmental Monitoring and Assessment, 193, 512. https://doi.org/10.1007/s10661-021-09234-5

Khan, M. A., Kumar, R., & Gupta, S. (2023). IoT-based smart air quality monitoring systems: A review of recent developments. Sustainable Computing: Informatics and Systems, 38, 100871. https://doi.org/10.1016/j.suscom.2023.100871

Kim, J., Park, Y., & Lee, K. (2022). Impact of sensor uncertainty on machine learning-based environmental prediction systems. IEEE Transactions on Instrumentation and Measurement, 71, 1–10. https://doi.org/10.1109/TIM.2022.3145678

Kumar, P., Morawska, L., Martani, C., & Biskos, G. (2022). The rise of low-cost sensing for managing air pollution in cities. Environment International, 164, 107253. https://doi.org/10.1016/j.envint.2022.107253

Li, Z., Zhao, Y., Sun, W., & Chen, Q. (2023). Time-series prediction of air quality using LSTM and ensemble learning methods. Environmental Modelling & Software, 162, 105634. https://doi.org/10.1016/j.envsoft.2023.105634

Liu, H., Wei, X., & Zhang, Q. (2023). Hybrid deep learning architecture for spatiotemporal air quality forecasting. Applied Soft Computing, 134, 110029. https://doi.org/10.1016/j.asoc.2023.110029

Maag, B., Zhou, Z., & Thiele, L. (2021). A survey on sensor calibration in air quality monitoring deployments. ACM Computing Surveys, 54(3), 1–36. https://doi.org/10.1145/3448304

Park, S., Kim, D., & Lee, H. (2021). Noise reduction techniques for low-cost environmental sensor data. IEEE Sensors Journal, 21(14), 15947–15956. https://doi.org/10.1109/JSEN.2021.3071123

Rahman, M. M., Islam, M. R., & Hossain, M. S. (2021). Edge-based real-time environmental monitoring using machine learning. Future Generation Computer Systems, 121, 87–97. https://doi.org/10.1016/j.future.2021.03.021

Singh, A., Gupta, R., & Sharma, N. (2022). Ensemble learning models for urban air quality prediction. Environmental Science and Pollution Research, 29, 52312–52325. https://doi.org/10.1007/s11356-022-19654-3

Spinelle, L., Gerboles, M., Villani, M. G., Aleixandre, M., & Bonavitacola, F. (2022). Evaluation of low-cost gas sensors for air quality monitoring applications. Atmospheric Measurement Techniques, 15(2), 475–489. https://doi.org/10.5194/amt-15-475-2022

Torres, J., Martinez, A., & Ruiz, D. (2021). Real-time environmental monitoring framework integrating IoT and AI. Computer Networks, 191, 107977. https://doi.org/10.1016/j.comnet.2021.107977

Wang, T., Li, M., & Chen, L. (2023). Performance comparison of regression algorithms for PM2.5 prediction. Atmospheric Pollution Research, 14(1), 101601. https://doi.org/10.1016/j.apr.2022.101601

World Health Organization. (2023). Global air quality guidelines update 2023. WHO Press.

Zhang, Y., Ding, A., Mao, H., & Fu, C. (2021). Machine learning approaches for air pollution prediction: A systematic review. Atmospheric Research, 250, 105348. https://doi.org/10.1016/j.atmosres.2020.105348

Zhou, X., Wang, S., & Liu, J. (2022). Real-time air quality prediction based on hybrid machine learning framework. IEEE Access, 10, 44321–44333. https://doi.org/10.1109/ACCESS.2022.3167890