Researches on the novel methodology of traffic flow prediction based on similarity

Similarity is one of the important characteristics of traffic flow time series, it reflects the same property of every two time series, and it is very significant to traffic flow prediction and traffic guidance. There are many prediction models based on similarity, but they neglect the direct impacts of historical data more or less. So, based on similarity and referring the idea of battery discharge effects, a new prediction model is proposed, it fully takes into account the impacts of historical data on current traffic flow, its description and characteristic analysis are given in detail. Finally, its performances is analyzed and compared with other models.