Research on algorithms of grey residual error model in highway traffic accidents forecast

By analyzing the procedure to establish Gray Residual Error Model(GREM), this paper classifies the factors that affect the forecast results with GREM, such as objects being amended, data being selected and methods for data pretreatment. Furthermore, several algorithms of GREM are presented, followed with highway traffic accidents forecast examples. Results indicate that errors will be greatly reduced by these methods in accident forecast, which has great randomicity as well as certain trend. Especially, the method that adopts residual error to establish GREM directly and recover once, has lower complexity, less error, more preferable effect and can be used in practice.