A novel hybrid model based on multi-objective Harris hawks optimization algorithm for daily PM2.5 and PM10 forecasting

High levels of air pollution can severely affect the living environment and even endanger the human lives. To reduce air pollution concentrations, and warn the public regarding the occurrence of hazardous air pollutants, an accurate and reliable air pollutant forecasting model must be designed. However, previous studies had numerous deficiencies, such as ignoring the importance of predictive stability and poor initial parameters; these deficiencies significantly affected the air pollution prediction performance. Therefore, in this study a novel hybrid model is proposed to address these issues. Powerful data preprocessing techniques are applied to decompose the original time series into different modes from low frequency to high frequency, and a new multi-objective Harris hawks optimization algorithm is developed to tune the parameters of the extreme learning machine (ELM) model with high forecasting accuracy and stability for prediction air pollution. The optimized ELM model is then used to predict a time-series of air pollution. Finally, a scientific and robust evaluation system with several error criteria, benchmark models, and experiments conducted using twelve air pollutant concentration time series from three cities in China is designed to assess the presented hybrid forecasting model. The experimental results indicate that the proposed hybrid model can achieve a more stable and higher predictive performance than other models, and its superior prediction ability may aid in developing effective plans for mitigating air pollutant emissions and preventing the health issues caused by air pollution. (C) 2020 Elsevier B.V. All rights reserved.