An Improved Traffic Flow Prediction Algorithm Based on TLBO and BP

BP has been shown to be effective in predicting traffic flow. However, BP neural net may reach the uncertainty of global minimum point. In this paper, an improved optimization algorithm is proposed to predict traffic flow, which is based on teaching-learning-based optimization and BP. The topological structure of BP is optimized by TLBO to search the minimum value of the objective function. In addition, a novel feedback phrase is proposed for convergence and accuracy. The simulation results show that the presented algorithm has a better predictive ability for traffic flow compared with some state-of-the-art approaches, and it has strong robustness and effectiveness.