Development of Optimal Combined Weighted Model for PM10 Forecasting

Forecasting air pollution is crucial for public health and pollution control, making statistical methods essential for this task. Therefore, developing an accurate and stable air pollution (AP) forecasting model is both urgent and necessary. Given the highly complex, volatile, and nonlinear nature of AP series, this research introduces a novel optimal combined weighted model (OCWF). The Artificial Bee Colony (ABC) algorithm is employed to optimize the weights of individual models, enhancing the proposed model's accuracy. This study proposes a combination forecasting method to develop an air quality forecast model for Petaling Jaya, Malaysia, utilizing meteorological data. The particulate matter (PM10) data from January 2003 to December 2022 are used to evaluate the proposed model. The model's performance is compared with that of the multiple linear regression (MLR),), Autoregressive Integrated Moving Average (ARIMA), artificial neural network (ANN), Long-Short-Term Memory (LSTM) and simple average model (SAM). The findings indicate that the proposed model exhibits superior generalization ability, more robust forecasting performance, and lower forecasting error.