PM2.5 Prediction using Deep Learning Models

Air quality monitoring and prediction are crucial in smart cities for public health, environmental sustainability, urban planning, emergency response, citizen empowerment, and data-driven decision making. By monitoring and predicting air quality, authorities can mitigate pollution sources, protect public health, and promote sustainable practices. It aids in informed urban planning, ensuring sensitive populations are not exposed to high pollution levels. Real-time monitoring enables timely responses during emergencies, protecting residents and responders. Citizens are empowered with access to accurate information, encouraging informed choices and collective action. Air quality data supports evidence-based decision making and resource allocation. Prioritizing air quality in smart cities enhances quality of life and fosters sustainable and healthy urban environments. In smart cities, we use air quality monitoring systems based on the Internet of Things (IoT) to control air pollution levels. Taking advantage of the good performance of deep neural networks in time-series prediction, these systems are upgraded to smart monitoring systems for improving performance and efficiency In this work, we propose to build deep learning models such as long short-term memory (LSTM), gated recurrent unit (GRU) and hybrid convolutional neural networks-long short-term memory (CNN-LSTM) to predict air pollution in the smart cities. First, we collect air data. Then, we adjust them through various preprocessing methods. Next, we deploy the different deep learning models. We train our models and test them using an air dataset. Finally, we evaluate prediction results of different models in terms of prediction accuracy using the coefficient of determination (R2), mean squared error (MSE) and root mean square error (RMSE) between the actual values and the predicted values.

I. Introduction

Over the past decades, as a result of human activities, industrialization and urbanization, pollution such as air pollution, water pollution, land pollution, etc, is becoming a critical issue. It has become a life-threatening factor in many countries of the world and a growing problem for the world’s population. The change in air quality will make the environment toxic and menace human being. It leads to severe problems, such as vulnerability to diseases, and increased death rate of humans, animals and plants. In smart city, where we talk about an interconnected world, we collect a huge amount of data from several domains in real time and we use it to improve quality of life. Air quality is an important topic and big issue in the smart city concept. According to the World Health Organization (WHO), more than seven million people die each year from this problem and more than 80% of the urban population lives in places where air quality exceeds WHO limits [24]. Hence, air quality monitoring and prediction are essential. Particulate matter PM_2.5 is one of the dangerous main air pollutants. It is a particulate with a diameter equal to 2.5 micrometres. This particulate reduced global life expectancy about 1.2–1.9 years in some polluted countries of Asia and Africa [25]. PM_2.5 has severe effects for human life, becoming the reason of about 3% of mortality from cardiopulmonary disease, 5% of mortality from cancer of the trachea, bronchus, and lung, and about 1% of mortality from acute respiratory infections in children under five year [26]. According to the following research [27], in 2015, PM_2.5 was the fifth-ranking mortality risk factor. Therefore, prevent or reduce consequences caused by air pollution is a crucial problem. If we forecast the trend in air quality changes, we can employ safeguards in advance to avoid imbalances in the ecosystem and also ensure suitable conditions for optimum life. Having information about air quality will encourage us to take protective measures; this can lead people to carry out their daily activities in less polluted places (by escaping from heavily polluted areas). However, analyzing data and providing smart solutions remains a difficult task. It is therefore essential to apply productive methods and techniques to more effectively analyze big data, convert the invisible into visible and extract the information hidden behind the data.