The invention provides an urban-
road traffic forecasting method based on multi-
source data combination. The method comprises the steps of firstly, extracting permanent residents' travel OD according to
cell-phone signaling data, and distributing the travel OD to an
urban road network to obtain distributional traffic flows of each road section; secondly, according to a bayonet
record, obtaining a total observed
traffic flow and an observed
traffic flow of frequently-used cars in a road section corresponding to the bayonet; thirdly, selecting a road section with observed traffic flows within the region, and building an equation of
linear regression which represents the time-varying correlation between the distributional traffic flows and the observed traffic flows in the road sections, according to the distributional traffic flows and the observed
traffic flow data; fourthly, according to the equation of
linear regression and a proportion of the frequently-used cars within the region, constituting a dynamic forecasting model of traffic flows of the road sections within the region; fifthly, as for the road sections without observed traffic flows in the region, inputting the distributional traffic flows of the road sections into the dynamic forecasting model to forecast the time-varying traffic flows of the road sections. The urban-
road traffic forecasting method based on multi-
source data combination has the advantages of providing convenience for obtaining information and conducting traffic forecasting work in multiple cities, and being low in costs and easy to operate.