Short-term traffic flow forecasting model of optimized BP neural network based on genetic algorithm

Focusing on the nonlinearity of traffic flow and easily running into local extremum of BP neural network (BPNN) in short-term traffic flow forecasting, the paper establishes the forecasting model based on BPNN and genetic algorithm (GA) which combines the stronger nonlinear approximation of BPNN and global search capability of GA. The genetic algorithm is introduced to search the optimal solutions of initial weight and threshold of BPNN, so as to improve the convergence and forecasting precision of network. The paper analyzes the chaotic characteristic of traffic flow, calculates embedding dimension and delay time, and reconstructs corresponding phase space which will be applied in the optimized model for short-term traffic flow forecasting. Simulation results show that the proposed method has better forecasting effect with high precision compared with traditional BP neural network.