Distributed state and unknown input estimation for freeway traffic flow models

The paper presents a joint state and unmeasured input estimation technique for a special class of nonlinear second order macroscopic traffic model. The freeway traffic system is transformed into a quasi and affine approximate Lin- ear Parameter Varying (LPV) form. Observer design technique is formulated using off-line optimization problem subjected to Linear Matrix Inequality (LMI) conditions, considering uncertain scheduling parameters. Furthermore, locally designed state estimators are interconnected to cover multiple freeway sections. The elaborated method is validated using real traffic measurements.