Related Experiment Video
Updated: Aug 14, 2025
Asthma Detection Research Based on Voice Signal Processing and Machine Learning
Published on: July 22, 2025
A performance comparison study on PM2.5 prediction at industrial areas using different training algorithms of
Pavithra Chinatamby1, Jegalakshimi Jewaratnam1
1Center for Separation Science & Technology (CSST), Department of Chemical Engineering, Faculty of Engineering, Universiti Malaya, 50603 Kuala Lumpur, Malaysia.
Abstract:
Presence of particulate matters with aerodynamic diameter of less than 2.5 μm (PM2.5) in the atmosphere is fast increasing in Malaysia due to industrialization and urbanization. Prolonged exposure of PM2.5 can cause serious health effects to human. This research is aimed to identify the most reliable model to predict the PM2.5 pollution using multi-layered feedforward-backpropagation neural network (FBNN). Air quality and meteorological data were collected from Department of Environment (DOE) Malaysia. Six different training algorithms consisting of thirteen various training functions were trained and compared. FBNN model with the highest coefficient correlation (R2) and lowest root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) were selected as the best performing model. Levenberg Marquardt (trainlm) is the best performing algorithms compared to other algorithms with R2 value of 0.9834 and the lowest error values for RMSE (2.3981), MAE (1.7843) and MAPE (0.1063).
Insights
Malaysia
Area of Science:
- Environmental Science
- Atmospheric Science
- Computational Science
Background:
- Particulate matter with aerodynamic diameter < 2.5 μm (PM2.5) is increasing in Malaysia due to industrialization and urbanization.
- Prolonged exposure to PM2.5 poses significant human health risks.
Purpose of the Study:
- To identify the most reliable model for predicting PM2.5 pollution in Malaysia.
- To evaluate the effectiveness of a multi-layered feedforward-backpropagation neural network (FBNN) for air quality forecasting.
Main Methods:
- Collected air quality and meteorological data from the Department of Environment (DOE) Malaysia.
- Trained and compared six different algorithms with thirteen training functions using FBNN.
- Selected the best performing model based on coefficient of correlation (R2) and error metrics (RMSE, MAE, MAPE).
Main Results:
- The Levenberg-Marquardt (trainlm) algorithm demonstrated superior performance among the tested algorithms.
- The best performing FBNN model achieved an R2 value of 0.9834.
- The model exhibited low error values: RMSE (2.3981), MAE (1.7843), and MAPE (0.1063).
Conclusions:
- The FBNN model, particularly with the Levenberg-Marquardt algorithm, is highly effective for predicting PM2.5 pollution in Malaysia.
- Accurate PM2.5 prediction models are crucial for mitigating health impacts associated with air pollution.
- This research provides a reliable computational approach for air quality management in rapidly developing regions.
