Traffic Flow Combination Forecasting Based on Grey Model and GRNN

摘要

This paper focuses on traffic flow forecasting which is an essential component in traffic control or route guidance system. A combination forecasting model called GM-GRNN based on GM(1,1) and GRNN is built for short-term traffic flow time series. The basic theory and features of General Regression Neural Network (GRNN) and its advantages are introduced. The weight of combination model is determined by optimal combination method. In the GRNN model, the number of input neurons and the value of smooth factor are determined by search method, and the forecasting process is single-step rolling forecasting. The results demonstrate that the GM-GRNN model with the advantage of all single models accurately fits the actual traffic flow, and has better performance than single model.