The Research Of The Application Of Neural Network In Short-Time Traffic Flow Forecasting

Road short-term traffic flow prediction is the core of the Intelligent Transportation System (ITS) and an important foundation of road traffic control and traffic flow guidance. The real-time and accurate prediction can not only alleviate the traffic congestion, but also improve road capacity and traffic safety. However, it is hard to establish a proper mathematical model because of its high uncertainty and non-linear character. Therefore, the short-term traffic flow forecasting model is established by adopting neural network in this paper. The main research work is as follows:Firstly, based on the study of BP neural network, this paper designs the short-term traffic flow forecasting model, detailedly analyses the process of mapping and the prediction performance of the model, and discusses the selection of the number of neurons in hidden layer and the preprocessing of data. What's more, the paper puts forward measures to improve network convergence speed.Secondly, because the BP algorithm is sensitive in setting initial network weights and thresholds and easy to fall into the local minimum, this paper uses the genetic algorithm, which is good at global search, to optimize the initial network weights and thresholds to improve the prediction performance.Finally, the model has been trained and simulated with the sample of the real short-term traffic flow data in the south second ring of Xi'an City. The simulation results show that the BP neural network forecasting model has simple algorithm structure without designing any mathematical model, and its accuracy can meets the requirements of short-term traffic flow. The improved genetic neural network forecasting model inherits the strong learning and training ability of BP neural network. It improves the forecast precision of the model at the same time.