Research On The Short-term Flow Forecast Model Of Urban Traffic Intersections

In the field of intelligent traffic control and vehicle guidance research,traffic flow forecasting technology in short-term traffic has significant research significance.Short-term traffic flow prediction usually refers to prediction based on acquired traffic data for the next 15 minutes,which has a great effect on improving real-time performance in traffic control and guidance.The key point in the intelligent transportation system is real-time,dynamic and accurate prediction of traffic flow to improve urban traffic management and operational efficiency.This is why short-term traffic prediction will become an important research content of the current intelligent transportation system.In addition,short-term traffic flow has a relatively short span of prediction time,and traffic data changes are sometimes not too strong.Various disturbance noises have a greater impact on traffic flow prediction,which undoubtedly leads to short-term traffic flow prediction.Development is currently very challenging.The paper first analyzes the main models and methods of short-term traffic flow in urban traffic at home and abroad,and analyzes the characteristics of traffic in short-term traffic.Based on the obtained traffic data,it hopes to find potential laws in traffic flow data.Sex.For this reason,the method of support vector regression was used to perform prediction research after analyzing the advantages and disadvantages of multiple regression prediction methods.Secondly,for the problem of how to find the optimal parameters in the method of support vector regression,the paper used the The particle swarm intelligence cluster optimization algorithm updates the parameters in the model,and improves the particle swarm itself may be trapped in the local optimal solution and the late oscillation;taking into account the use of improved particle swarm optimization support vector regression Short-term traffic forecasting can indeed produce good forecasting results under normal conditions,but some unexpected events occur frequently in traffic.In this case,the effect of the model will be greatly reduced.Based on this,using the relevance existing in the road network,a temporal-spatial association prediction method is proposed,which is integrated with the improved particle swarm optimization support vector regression,and the weightvalue between the two can be iteratively updated using the BP neural network.The characteristics form an intelligent short-term traffic flow prediction model that can be self-learned.