Reseatch On Traffic Flow Prediction Technology For Intelligent Traffic Junction

With the rapid development of society, the traffic capacity is related to the development of cities economy and the advance of the urbanization process. It has become the focus of common concern in today's to protect the city traffic safety and finish traffic intelligent. Intelligent Transport System (ITS) is an efficient method to alleviate urban traffic pressure and solve the problem of traffic congestion at present. Collection and prediction of short time traffic flow is an important part of ITS. Timely and accurate squeezing traffic information helps to ensure the smooth flow of urban roads. At the same time, it provides reliable and accurate data support for the induction and control of urban traffic. It is also the key and foundation to relieve traffic congestion and avoid traffic accidents. To this end, this paper mainly focuses on the research of traffic flow prediction technology for intelligent traffic junction. Aiming at the problem of short term traffic flow forecasting at the junction, this paper presents the traffic information collection technology based on video image analysis and the traffic flow prediction algorithm based on combination model. It also designs and implements traffic flow collection and prediction system, then uses the system to analyze and forecast the traffic flow at the intersection.The traffic information collection technology based on video image analysis firstly obtains the background image by the average background model, sets up the virtual detection coil, and then uses the background subtraction method to complete the collection of traffic data. It counts the number of cars according to the change of gray value in the virtual coil.The simulation results show that the video traffic flow information collection technology based on the background subtraction method is more accurate than the frame difference method.The traffic flow prediction algorithm based on combination model clusters the collected traffic data by the traditional K-means algorithm,divides the whole data domain into several sequences according to the time by introducing the time base line, and then uses Extreme Learning Machine (ELM) to predict each segmentation sequence. The simulation results show that, compared with BP neural network and ELM, the modeling time of the traffic flow prediction algorithm based on combination model is shorter, the error is smaller, and the model is more credible.Traffic flow collection and prediction system includes traffic flow collection, traffic flow collection data management, traffic forecast and traffic forecast data management function. The system can help alleviate the pressure of urban traffic congestion, and has important practical significance.