Research On Prediction Of Short-time Traffic Flow Based On Neural Networks Optimized By Simulated Annealing Algorithm

Nowadays the intelligent transportation technology play more and more important role in the daily organization of the traffic system.The visualization and dynamic of the initiative traffic organization provided by intelligent transportation technology often give theoretical guidance and technical support for the decision of the management of the transportation.The prediction of traffic flow is one kind of the intelligent transportation technologies.It could help to offer the advanced data support for the technique of traffic flow induction and traffic distribution and the optimization of driving path in intelligent transportation technology and the selection of the path made by drivers.So the prediction of short-time traffic flow become a popularity issue in the management of transportation.The issue which is based on the fundamental and technical features of the BP network to analyze the defection of BP network which easy to fall into the local optimum during the phase of training and then combine with the ability of simulated annealing algorithm which could find the best solution from the situation as a whole to get the SA-BP neural network optimized by simulated annealing algorithm.On the basis of the optimization the issue research the method to predict the short-term traffic flow used by SA-BP neural network and then contrast and verify of the issue explain the effect of the prediction after the optimization is better than the passed by the example.The mainly contents of the issue are as follow: Firstly,discussed the background and the significance of the issue and made focus on the research of the prediction of short-time traffic flow technique and the optimization of SA-BP neural network.Secondly,introduced the theories and methods of the prediction of short-time traffic flow technique.Introduced the methods of data processed which will be used by the prediction.Made some evaluation indexes for the research.Thirdly,made the choice of the method of short-term traffic flow with BP neural network,and analyzed the operating principle of BP neural network and discuss the its defection as well as the simulated annealing algorithm and its advantage.Fourthly,made a mechanism which use the simulated annealing algorithm to optimize the BP neural network and get the SA-BP neural network by cell array matrix from Matlab.Suggesting a memory box function to optimize the results of the prediction form SA-BP neural network.Fifthly,made the research of the prediction for short-term traffic flow base on the SA-BP neural network and analyze the training and the prediction process and summarize the flow path of the prediction process.Finally,the issue uses an example of the section of the road from Shenzhen to simulate with MATLAB by SA-BP neural network and make the contrast with BP neural network,proving the SA-BP neural network could help to solve the defection of BP neural network and could have more accuracy results of prediction for short-term traffic flow.