Font Size: a A A

Neural Network Predictive Control For Freeway Traffic System

Posted on:2007-01-05Degree:MasterType:Thesis
Country:ChinaCandidate:D L WangFull Text:PDF
GTID:2178360185987119Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Ramp metering is regarded as one of the best control strategies for freeway traffic system and is widely used. Through analyses of existing ramp metering algorithms, intelligent control strategy is considered as an effective approach.On the basis of macroscopic dynamic traffic flow model which is frequently used in traffic control, Radial Basis Function (RBF) neural network is designed. Considering the robustness of predictive control, then, Genetic-Algorithms (GA)-based entrance ramp predictive control is achieved.Firstly, because of nonlinearity and uncertainty of freeway traffic flow, an accurate mathematics model can not be obtained. Therefore, RBF neural network model to predict the future behaviors of freeway traffic flow is designed based on MATLAB neural network toolbox. According to the results of approximation and generalization capability, RBF neural network model can accurately depict the behavior of freeway traffic flow and capture the inherent time-space interrelations.Then, predictive control strategy with features of model predicting, horizon receding and feedback correcting, is employed to overcome characteristics of nonlinearity, time delay, uncertainty and strong disturbance existing in traffic systems. The GA technique is robust under the presence of nonlinear structures in the cost functions and constrains, thus is used for optimization in the proposed controller.Simulation results, comparing with no control case, indicate the effectiveness of GA-based on-ramp predictive control. Furthermore, compared with the optimal control algorithm, the proposed controller has the advantages of stronger robustness...
Keywords/Search Tags:freeway, ramp metering, Radial basis function (RBF) neural network, predictive control, Genetic algorithms
PDF Full Text Request
Related items