Study On Railway Freight Arrival Time Prediction Based On Random Forest Algorithm

With the introduction of more high-tech into the railway industry,the efficiency and accuracy of the railway industry have been improved,and the transportation efficiency of the railway industry has been improved to a certain extent.In today's freight transport market,the arrival time of goods has become more and more the focus of enterprises and customers in choosing the mode of goods transportation.In order to cope with the competition in the transportation market,especially with the increasing transportation volume of high-value goods such as white goods,under the existing prediction model of the arrival time of goods,due to the huge amount of message data of goods transportation,the road network lines In addition,it is impossible to accurately predict the arrival time of different types of goods and the arrival time of different types of freight,which can not meet the requirements of different types of railway transportation In terms of customer demand and railway dispatching,the time required for directional train flow passing through technical operation at the current station is an urgent problem to be solved.This paper mainly studies the following four aspects:(1)This paper inquires the related papers on the prediction of railway freight arrival time,analyzes the factors that may affect the prediction of railway freight arrival time and the difficulties of railway freight arrival time prediction.Due to the time limit of railway freight transportation is the longest transportation time promised by the freight department to customers,but the actual arrival time of goods often causes more delays,which can not guarantee the interests of both customers and the railway industry.(2)Based on the comparison of previous scholars' prediction of railway freight arrival time,this paper briefly describes each time prediction method,analyzes and points out the imperfections of each prediction method,and puts forward a new prediction method of railway freight arrival time,(3)The construction of random forest algorithm based on big data platform is completed.According to a large number of vehicle arrival message information,some feature vectors which may affect the arrival time are extracted as independent variables,A single regression decision tree is established.The whole vector space is divided into several partitions,and the complete random forest is constructed.The decision tree is trained by learning.After inputting the independent variables,the set of prediction time of the corresponding decision tree is obtained,and the average value is calculated to obtain the final freight arrival time prediction value.On this basis,the error is analyzed by comparing the predicted time with the actual time ?(4)This paper reviews the whole random forest algorithm,analyzes the problems and deficiencies in the actual prediction,and puts forward the possible solutions and the future optimization direction.