Research on Short-term Traffic Forecast Algorithm based on Cloud Model

Abstract

Short-term traffic flow is difficult to predict because of high uncertainty. This paper proposes a short-term traffic forecast algorithm based on cloud similarity. By taking advantage of quantitative and qualitative cloud model mutual conversion function and traffic flow predictability, the historical traffic data can be processed with cloud transformation. Set the current traffic cloud as a standard, traverse the historical traffic cloud to find the best traffic flow period which is with best similarities to the current traffic clouds. Set the future short-term traffic flow of this very period of time as the prediction result of the current period of time. Experiments show that the average prediction error was 3.25 (vehicles) and the prediction error distribution probability was 0.29.

DOI : http://dx.doi.org/10.11591/telkomnika.v12i3.4655