Macroscopic fundamental diagrams (MFD) can be used to control the boundary of saturated traffic networks at the macroscopic level, and connected-vehicle network provides new technical means for real-time access to traffic parameters. Dividing sub-areas of homogeneous traffic networks is a prerequisite for effective boundary control. At present, most sub-areas of homogeneous traffic networks can get single-layer boundary, but queuing overflow may occur at the boundary under single-layer boundary control. Therefore, it is necessary to dynamically divide the homogeneous region into the same core multi-layer boundary, which lay a foundation for the follow-up implementation of multi-layer boundary control. A dynamic partitioning method based on MFD for the control boundary of over-saturated homogeneous connected-vehicle network is proposed in this study. This method utilises set theory. A traffic networks model based on road links is established and defined as the minimum unit of the traffic networks according to MFD correlation theory and historical data. A minimum unit MFD database of the traffic networks is also created. The weighted density and the weighted flow of the minimum unit of each traffic networks can be calculated based on the trajectory of networked vehicles in connected-vehicle networking. The smallest unit in the traffic networks with the maximum weighted traffic density is the key node of the network according to the MFD database. The central key node is extended to the peripheral nodes until the boundary of the congested area of the saturated traffic networks is determined. The control boundary of the multilayer traffic networks, including the congestion area, transition zone, and normal region, is established. A simulation model of the actual traffic networks in Tianhe District of Guangzhou is established using Vissim traffic simulation software, and the MFD database is set up. Another development method is realised by using C# language. The multi-layer control boundary of the traffic networks is obtained and compared with that in actual conditions of the traffic networks on a Baidu map to verify the validity of the proposed method. © 2019, Gioacchino Onorati Editore. All Rights Reserved.
