Unscented stochastic model predictive perimeter control of uncertain two-regions urban traffic

Abstract

In this article, Stochastic Model Predictive Control (SMPC) is employed for optimal perimeter control of traffic flow with uncertain Macroscopic Fundamental Diagram (MFD), traffic accumulation and traffic demand of two regions. Two regions urban traffic networks are described through the MFD. The MFD is a fundamental relation between average flow (production) and density (accumulation) in urban regions. Although the MFD is often assumed as a simple deterministic curve, possible heterogeneity in urban regions results in large scattering of the MFD pattern. Traffic accumulation is considered uncertain due to limited sources of measurements. Moreover, traffic demand is based on the stochastic nature of drivers. The stochastic uncertainty is modeled through appropriate probability distribution functions for MFD, traffic accumulation and demand. Simulation results show the superiority of the proposed method compared to deterministic MPC in the presence of model mismatch.