Prediction of particulate matter concentration in Chengdu based on improved differential evolution algorithm and BP neural network model

Abstract

Based on the establishment of improved BP neural network model, this paper has carried on the research to the PM by collecting the particle concentration and the related meteorological data for prediction. In this paper, the improved difference evolution algorithm is used to optimize the weights and thresholds of the BP neural network model (IDE-BPNN) for the traditional BP neural network model, which is too dependent on the initial value, the convergence rate is slow and easy to fall into the local minimum. The improved model is compared with the traditional BP neural network model and the other five optimized BP neural network models. The root mean square error (RMSE), the mean absolute error (MAE) and the mean absolute percentage error (MAPE) are smaller than those of other models, and the index of agreement (IA) of IDE-BPNN is the highest, the mean bias error (MBE) of IDE-BPNN tends to be zero, and the IDE-BPNN model performs better.