Traffic Flow Forecasting Algorithm Using Simulated Annealing Genetic BP Network

Genetic back propagation (BP) neural network is fast, quick, steady in forecasting of traffic flow, and the result has lowly error ability. But it can easily cause premature convergence, and usually the solution we got is local optimal solution. For overcoming those drawbacks of Genetic BP neural network, we add Simulated Annealing Algorithm to the processing of GA, using the ability of Annealing Algorithm that can get rid of local optimum to restrain the premature of GA and reduce the selection pressure. The results of simulation experiment results of the cross road´s short-term traffic flow forecasting show that the algorithm can not only overcome the premature of Genetic Algorithm but also can increase its robustness, and at the same time reduce iterative numbers and the error of traffic flow forecasting, raise the forecast precision.