Modeling and fault-tolerant control of large urban traffic networks

This work deals with the modeling and design of fault tolerant traffic control systems for urban transportation networks. The dynamics of queue length in front of the intersections are captured by a nonlinear state space model where a limit-reset function is introduced to maintain the non-negativity of the queues for traffic conditions ranging from undersaturation to oversaturation. A fault-tolerant control system is then synthesized to maintains system integrity (stability and vehicle flow regulation) under partial system failure brought about by local faults such as sensor/actuator failure, vehicle accidents, severing of communication links, etc. It is shown that the fault-tolerant conditions are rather mild and can readily be satisfied by a properly designed decentralized PI-type controller. A detailed case study is also included in this work to demonstrate the properties of a fault tolerant traffic network