Short-term PM2.5 prediction based on variational mode decomposition and machine learning methods

Accurate prediction of PM2.5 concentration can effectively avoid the harm caused by air pollution to human body.The existing PM2.5 prediction models generally have the problems of low accuracy and long prediction period, for this reason, this paper proposes a combined model based on variational mode decomposition and machine learning methods(least squares support vector machine, echo state network and extreme learning machine) to improve the accuracy of short-term PM2.5 prediction, and establishes benchmark models and hybrid models for comparison and analysis, the performance of each model was evaluated by two evaluation metrics separately.The experimental results show that the model has the optimal prediction performance.