In order to effectively improve the precision of short-term traffic flow prediction

a short-term traffic flow prediction model based on self-variation fireworks algorithm（ FWA） is proposed. The self-variation fireworks algorithm adaptively selects the variation fireworks using the adaptive fireworks mutation operator in the optimization process of the back propagation（ BP） neural network according to the algorithm. Random mapping rules is adopt for fireworks outside the border to increase population diversity. The constructed prediction model can accurately predict the nonlinearity and complexity of short-term traffic flow and accelerate the optimal solution. In 2 or 5 min traffic flow forecasts

the proposed model shows better performance than the BP neural network model

particle swarm optimization（ PSO） BP neural network and fireworks algorithm BP neural network model.