Atmospheric PM2.5 Prediction Using DeepAR Optimized by Sparrow Search Algorithm with Opposition-Based and Fitness-Based Learning

There is an important significance for human health in predicting atmospheric concentration precisely. However, due to the complexity and influence of contingency, prediction a challenging topic. In this paper, we propose novel hybrid learning method make point interval predictions PM2.5 simultaneously. Firstly, optimize Sparrow Search Algorithm (SSA) by opposition-based learning, fitness-based Lévy flight. The experiments show that improved (FOSSA) outperforms SSA-based algorithms. addition, employed initial weights probabilistic forecasting model with autoregressive recurrent network (DeepAR). Then, FOSSA–DeepAR utilized achieve Beijing, China. performance compared other models single DeepAR model. Furthermore, hourly data O3 Taian China, China are used verify effectiveness robustness proposed method. Finally, empirical results illustrate can more efficient accurate both prediction.