关键词: 交通流预测, 广义神经网络, 并行计算

Abstract: Real-time and accurate traffic flow prediction is crucial to the development of traffic flow guidance system.In order to solve the problem that BP network has the disadvantages of slow convergence speed and target function getting into local minimal value，this paper employs general neural network composed of intelligent neuron to forecast traffic flow.Meanwhile，a new parallel training algorithm based on training set decomposition is presented.This algorithm uses a dynamic communication profile.Using this algorithm to train general neural network and doing experiments with real traffic flow data of Dalian，experiment results show that the new algorithm improves both convergent speed and forecasting precision，and can meet practical requirement.

Key words: traffic flow prediction, generalized neural network, parallel computing