Algorithm for the Traffic Flow Prediction Based on Improved Non-Parametric Regression Method
|School||Huazhong University of Science and Technology|
|Keywords||Short-term traffic flow prediction Nonparametric regression Bayesian classifier Varying the value of K algorithm|
Increasingly severe traffic problems encountered by the world 's major cities , At the same time , the existing intelligent transportation systems play an increasingly significant role in alleviating traffic problems , intelligent transportation systems has received widespread attention . Traffic flow control and guidance system acts directly on the road to alleviate traffic congestion problems in the most direct . Make the traffic flow control and guidance systems can be a good role to play , accurate , real-time traffic flow forecast information is crucial . Directly by the road network and vehicle transport system has greatly nonlinear , complexity and uncertainty . Existing traffic flow forecasting methods in predicting whether real-time or in the accuracy of forecast and actual demand , there are some gaps, can not meet the demand for intelligent transportation systems at the present stage . This paper examines several common traffic flow forecasting methods , nonparametric regression method proposed some improvements. Specifically , mainly of traditional non-parametric regression prediction method proposed two improvements : Bayesian network methods to classify the traffic flow status , classification, to seek to reduce the non- parametric regression method neighbor subset search time , which greatly reduces the computation time to improve the real-time nature of the algorithm ; Bayesian network to classify the traffic flow state , using different values ??of K , K-nearest neighbor nonparametric regression algorithm to predict different types of traffic flow state to improve algorithms accuracy . Nonparametric regression algorithm improved simulation proved the effectiveness of the algorithm . The simulation results show that the improved non- parameter regression method than the traditional method improved algorithm for real-time algorithm accuracy . Improved algorithm helps make intelligent transportation systems play a role , to cope with the pressure of traffic problems , reduce traffic problems brought huge economic losses .