Combined control predictive model and method for stochastic traffic flow
DONG Wei
Abstract：In order to solve the problem of time variability, randomness and uncertainty in the traffic flow forecasting, a new combined control predictive algorithm was proposed. The proposed method divided the traffic flow into uniform part and random part with the multi-scale wavelet analysis method. The combined control predictive model was established according to the dynamic change of states. The uniform and random sequences of traffic flow were predicted with the support vector machine regression and Markov chain method, respectively. Based on the quasi dynamic programming method, the optimal control vector and the corresponding control matrix were calculated. Hence, the optimal prediction results could be obtained through the dynamic selection of training data. Furthermore, the experimental verification and comparison were implemented with the actual data. The results show that the average relative error and mean square deviation obtained with the proposed method are 74% and 85% lower than those of other methods, and the obtained equality coefficient is 6% higher than that of other methods. Therefore, the feasibility and effectiveness of the proposed method can be verified.
Keywords：traffic flow forecastingmulti-scale waveletsupport vector machine regressionMarkov chainquasi dynamic programmingoptimal control vectoroptimal control matrixcombined control prediction
Publication Date：2018-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 88-93 )