Short-Term PM2.5 Forecasting Using Exponential Smoothing Method: A Comparative Analysis.

Area of Science:

• Environmental Science
• Data Science
• Sensor Technology

Background:

• Air pollution, particularly fine particulate matter (PM2.5), poses a significant global environmental and health challenge.
• Accurate and economical monitoring and forecasting of PM2.5 concentrations are crucial for public health and environmental management.
• The Internet of Things (IoT) offers potential for developing agile and cost-effective air quality monitoring systems.

Purpose of the Study:

• To develop and evaluate an economical air quality forecasting system using IoT devices.
• To implement and test a novel forecasting method based on exponential smoothing with drift for PM2.5 prediction.
• To compare the performance of the proposed model against existing forecasting models in terms of accuracy and computation time.

Main Methods:

• Utilized a historical data-based approach for PM2.5 forecasting.
• Developed a forecasting model employing exponential smoothing with drift.
• Conducted experiments using real-time PM2.5 data from 132 IoT monitoring stations in the Taichung region, Taiwan.

Main Results:

• The proposed model achieved a low error rate of 0.16 μg/m³ for PM2.5 forecasting across 132 stations.
• The model demonstrated an acceptable computation time of 30 seconds for forecasting.
• Further evaluation showed that 90% of stations had prediction errors under 1.5 μg/m³ for 3-hour ahead forecasts.

Conclusions:

• The developed IoT-based system provides an accurate and efficient method for PM2.5 air quality forecasting.
• Exponential smoothing with drift is a viable technique for real-time PM2.5 prediction with IoT deployments.
• The system offers a cost-effective solution for monitoring and forecasting air quality, contributing to environmental management strategies.