Research On Traffic Congestion Prediction Based On Hadoop Big Data Platform

Traffic congestion is becoming a severe phenomenon in China due to the urbanization.The Intelligent Traffic System,which improves the accuracy of the traffic flow information timely,can help the Traffic Flow Guidance Systems to operate better and effectively reduce the traffic congestion.With the development of the traffic information collection system,the volumes of traffic data are increasing exponentially,resulting in the ‘big data' era.Traffic big data brings opportunities and challenges to the intelligent transportation field.Massive traffic data imply valuable information.However,the traditional traffic data processing technology cannot be applied to traffic big data.The use of big data technology to mine massive traffic data and extract valuable information to solve real-life traffic problems is an important research direction in the current traffic field.This paper analyzes the characteristics and existing problems of traffic big data,and uses Hadoop technology to build a big data platform to handle traffic big data.The preprocessing work of traffic big data is performed on the Hadoop big data platform with instance data.Based on this,a combination of machine learning algorithms is used to predict the traffic conditions and the prediction results are accurate and timely.Hadoop can meet the needs of actual traffic management.In this article,I mainly carried out the following research:This paper analyzes the characteristics and existing problems of traffic big data,and uses Hadoop technology to build a big data platform to handle traffic big data.The preprocessing work of traffic big data is performed on the Hadoop big data platform with instance samples.Based on this,a combination of machine learning algorithms is used to predict the traffic conditions and improve the accuracy of forecast results in time.Hadoop can meet the needs of actual traffic management.In this article,I mainly carried out the following research:First,the research focuses on the traffic big data technology and the methods of forecast the traffic congestion nowadays,and illustrates the characteristics of traffic big data and the problems it brings.Also,it introduces the Hadoop system,which mainly consists of three components,and explains the principles and advantages of the data processing within the system,as the basis for research.Secondly,there is an essential concern on the quality of data mining in the processing stage due to the necessity and complexity as well as the disadvantage of the traditional data processing technology.Moreover,it provided a practice and implication of traffic data processing by using the Hadoop big data platform.Thirdly,this paper defines and classifies the traffic congestion according to the previous research.It introduces the advantages and disadvantages of common traffic congestion forecasting algorithms,and studied the applicability of the algorithm under the framework of MapReduce computing,and obtained suitable algorithms that can complement the Hadoop technology advantages.Combined with traffic congestion forecasting requirements,the algorithm is adjusted;the algorithm is designed and implemented in parallel with MapReduce.Theoretically,it can improve the algorithm's accuracy and computational efficiency.Finally,it implies to set up a pseudo-distributed Hadoop big data platform to simulate traffic jam predictions with instance traffic big data.The experiment was compared with conventional traffic jam prediction algorithms,and the error of experimental results was analyzed to prove the feasibility and practicality of Hadoop processing traffic data for traffic jam prediction.