HDLP: air quality modeling with hybrid deep learning approaches and particle swam optimization

Predicting air pollution in cities has become an important tool for preventing its negative impacts. Therefore, citizens should be aware of air quality level, especially for individuals suffering from diseases caused by air pollutants. Collective efforts from researchers, environmental institutions, governments, industrial companies, and policy makers are shaping the future of the Air Quality Index (AQI) to effectively address severe air pollution in urban areas. Many air quality prediction models have been introduced in the literature; modern advances in deep learning techniques are promising more precise prediction results and data integration. The aim of this paper is to review methods, contributions, findings, limitations, and gaps in predicting air quality index and PM2.5 concentrations using a hybrid deep learning approach. A literature review led researchers to propose a Hybrid Deep Learning model with Particle Swarm Optimization (HDLP) which combines CNN, LSTM, and PSO. Algorithm, first Discrete Wavelet Transform (DWT) is used to solve the air pollution signal, then it is fed to CNN-LSTM neural network for PSO optimization to get the final prediction result. Optimized parameters input models are trained on the original data. In addition to the beneficial assessment cycle, to outperform current models.