Research on Earning Warning of Railway Freight Based on Support Vector Regression (SVR)

Abstract

The traditional railway freight early warning adopted independent indicator thresholds discrimination. But it would lead to inaccurate results warnings, it may alarms when abnormal events occur instead of early warning. This paper proposes a railway freight early warning model, which apply support vector regression (SVR) to rail freight early warning. this model could find their variation, sum up the formation of abnormal alarm mod by analyzing historical indicators, including of number of booking railway freight vehicles, loading vehicles, freight volume and freight turnover quantity. Early warning signals will be given when an exception occurs at the beginning of freight. Early analysis and processing potential risk are as to stabilize railway freight market. Experimental result shows a high prediction accuracy of this model for the railway freight market.