SPARSE MODELLING FOR LOCATION DEPENDENT FUNDAMENTAL DIAGRAM WITH COMPLETE VEHICLE TRAJECTORY DATASET

Abstract

 The Fundamental Diagram (FD) plays an important role in the performance evaluation of road networks. As FDs are in principal different among locations, it is desirable to estimate FDs at arbitrary locations based on actual data. Given a complete vehicle trajectory dataset, we propose a method to estimate each FD at each location. For this purpose, it is necessary to identify the sections that follow the different FDs each other. Using sparse modeling, we formulate a model that can simultaneously perform section identification and FD estimation. This enables us to determine a statistically and/or practically suitable number of sections without requiring a priori knowledge of road performance. The proposed method is applied to the actual data on the Hanshin Expressway and the validity of the estimation results is demonstrated.