Parameter Identification and Model Validation of a Macroscopic Traffic Model. Issue 1 (July 2017)

• Record Type: Journal Article
• Authors: Thonhofer, Elvira; Fuhrmann, Martin; Jakubek, Stefan
• Is Part Of: IFAC-PapersOnLine. Volume 50:Issue 1(2017)
• Page Start: 12835
• Page End: 12840
• Publication Date: 2017-07
• Subjects: Traffic model -- Optimization -- Parameter estimation -- Sensitvity analysis -- Validation; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2017.08.1933
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: Adaptive traffic control algorithms require an underlying traffic model. A suitable macroscopic traffic model has previously been developed by the authors. In this work the proposed method of parameter identification and the traffic model are validated with real live traffic data PeMS (Performance Measurement System) provided by the California Institute of Transportation. The proposed method of parameter identification utilizes a genetic algorithm (GA) where the difference between measurement data and simulation data are minimized. Parameter sensitivity and identifiability are investigated via the Fisher Information Matrix. The macroscopic traffic model based on the identified parameters is used to simulate traffic for a given time period at the test field. Results are presented and cross-validated with both the provided data of the PeMS data base and simulation results of the stochastic cell transmission model.