Short-Term Traffic Flow Prediction Method Based on WT-IGWO-ELM

Abstract

Accurate prediction of short-term traffic flow provides crucial data support for the stable operation of intelligent transportation systems. For this issue, this paper proposes a short-term traffic flow prediction method based on WT-IGWO-ELM. The algorithm uses the Wavelet Transform (WT) method to denoise the traffic flow data in advance, which improves the data quality of the dataset. Then, the IGWO algorithm, which integrates the initial population based on Sine chaotic map and reverse learning strategy, the adjustment of nonlinear convergence factor and the introduction of dynamic weights, is used to avoid local optimality more effectively, speed up the convergence speed, and improve the solution accuracy. Finally, the improved grey wolf optimizer (IGWO) was used to update the optimal parameters of the ELM prediction model, and the average relative error of the prediction of the WT-IGWO-ELM model was verified by comparison experiments compared with those of ELM and WT. -ELM, GWO-ELM, WT-GWO-ELM and IGWO-ELM decreased by 96.6625%, 95.5972%, 87.9447%, 79.5021%, 72.0571%, respectively, and its prediction effect was much better than ELM, WT-ELM, GWO-ELM, WT-GWO-ELM and IGWO-ELM methods have high prediction performance and accuracy in short-term traffic flow prediction.