Constrained model-free adaptive perimeter control and route guidance for large-scale urban networks under aggregate unmeasured disturbance effects

Urban traffic networks are increasingly exposed to time-varying demand, measurement uncertainty, and macroscopic fundamental diagram (MFD) scatter, which complicate the design of perimeter control (PC) and route guidance (RG) strategies. Conventional model-dependent methods usually require calibrated MFDs, demand forecasts, and relatively stable operating conditions; their performance may therefore deteriorate when unmeasured demand variations, sensing errors, or MFD scatter affect regional traffic-state evolution. This study develops a constrained model-free adaptive control (cMFAC) strategy that coordinates PC and RG for large-scale multi-region urban traffic systems using online input-output data. A pseudo-Jacobian disturbance-effect term is incorporated into the compact-form dynamic linearization model to compensate for the aggregate effect of unmeasured disturbances on regional accumulations. This term provides online compensation for disturbance-induced increments in regional accumulations, without requiring identification of the physical sources of the disturbances. Feasible ranges of perimeter release ratios, route guidance proportions, and regional accumulations are enforced through a constrained quadratic programming formulation. Numerical simulations on a six-region network in Zhengzhou are used to compare the proposed method with representative benchmarks under stochastic measurement uncertainty and MFD-scatter uncertainty. The results show improved accumulation regulation and network-level efficiency in the simulated scenarios.