A prediction model for traffic congestion in complex urban road networks

Some traditional Markov prediction models such as single variable model can only solve traffic on a single time series prediction problem. The first-order model only considers the influence between adjacent time point data, but the prediction precision of the higher-order multivariable Markov model needs to be improved. These models are difficult to solve traffic congestion prediction problem in complex urban road networks. This paper proposes a add adjustment term to the higher-order multivariable Markov model (AAT-HO3M) with convergence and estimation of the parameters. This model is applied in traffic congestion prediction. The results of the predictions illustrate that the prediction precisions of AAT-HO3M are higher than the traditional higher-order multivariable Markov model and improved multivariable Markov model, and the time overheads of AAT-HO3M are less than the improved multivariable Markov model.