A wavelet enhanced approach with ensemble based deep learning approach to detect air pollution

Air pollution is one of the most serious environmental problems worldwide. Especially, PM2.5 cannot be blocked by filtration system of respiratory tract to cause serious threat to public health. Considering the serious impact of air pollution on human health and life, air pollutant concentration forecast has been drawn great attention because it can provide people with information of air quality. In order to improve the accuracy of air pollutant concentration forecast, this study proposes a new Ensemble Empirical Mode Decomposition (EEMD) multi-step forecasting model is proposed. Initially, input data are decomposed by enhanced empirical wavelet transform (EEWT) to increase the dimensionality of the data. Next, EEWT-NLSTM, EEWT-DLSTM and EEWT-Bi-LSTM models are constructed using wavelet decomposition results and Nested Long short-term memory neural network (NLSTM), Deep Long short-term memory neural network (DLSTM), and Bi-directional long-short term memory neural network (Bi-LSTM), respectively, followed by comparison and contrast of the prediction results. Three single prediction models are incorporated into combined weighted forecasting model by weight assignment, it improves the prediction accuracy. To accurately assess effectiveness of the experimental model, three evaluation metrics, Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE), are used in this paper to quantitatively evaluate prediction performance of proposed model. For the prediction of PM2.5, the proposed model achieved a Mean Absolute Error (MAE) of 0.047428.