A New Hybrid Forecasting Model Based on Dual Series Decomposition with Long-Term Short-Term Memory

In recent years, ozone (O3) has gradually become the primary pollutant plaguing urban air quality. Accurate and efficient prediction is of great significance to prevention control pollution. The quality monitoring network provides multisource concentration data for prediction, but based on still faces challenges each station’s series data. Aiming at problems low accuracy computational efficiency in traditional atmospheric using dual decomposition was proposed by variational mode (VMD), ensemble empirical (EEMD), long short-term memory (LSTM). First, historical Nanjing stations decomposed VMD, then EEMD algorithm applied residual VMD obtain several characteristic intrinsic function (IMF) components; IMF component trained LSTM result component, final can be obtained linear superposition. method achieved best results with R2 = 99%, MSE 5.38, MAE 4.54, MAPE 3.12. Because strong adaptive learning ability good function, it advantage long-term data, are more accurate. According superior baseline models terms statistical metrics. As a result, hybrid serve as reliable model forecasting.