Macroscopic Model Free Adaptive Perimeter Control Methods For Urban Traffic Systems

In this thesis,some research has been done in the fields of macroscopic perimeter control(PC)and model free adaptive control(MFAC)of urban traffic network.The scientific significance of this work is that the dependence on macroscopic fundamental diagram(MFD)or other mathematical models of urban traffic systems in the study of PC is avoided.Instead,only the input and output(I/O)data of the urban traffic system are needed to design the perimeter controller.The main research contents and contributions of this work are summarized as follows.1.According to the characteristics of large-scale urban traffic system,such as large scale,difficulties in modeling,strong coupling,nonlinearity and randomness,a decentralized estimation and decentralized model free adaptive control(DED-MFAC)method is proposed and applied to PC of urban traffic system.Firstly,the complexconnected large-scale urban traffic system is decomposed into several regions,where each region is regarded as a subsystem,and the traffic flows of the vehicles among the regions are seen as the influences of interconnection among the subsystems.Secondly,the compact form and partial form dynamic linearization data models are established for each subsystem with interconnected influences,respectively.Finally,the pseudo gradient(PG)and pseudo partitioned gradient(PPG)estimation and perimeter controller design for each subsystem are proposed respectively.The urban traffic model is no longer required,and only the measurable interactions among the subsystems are needed.2.Considering the physical constraints of both control input and system output and the unavailable couplings among different regions due to the high cost in urban traffic systems,a novel centralized method called model free adaptive control with input and output constraints(IOC-MFAC)for unknown multiple-input and multiple-output(MIMO)nonlinear systems is proposed,and it is applied to solve both PC and route guidance(RG)problems simultaneously for the multi-region urban traffic system(MRUTS).Through a series of simulation experiments and comparative verification,the effectiveness and superiority of the proposed IOC-MFAC method are verified.3.Aiming at the prediction and control problems of the MRUTS,a data driven method called constrained model free adaptive predictive control(c MFAPC)for unknown nonlinear MIMO systems is proposed for PC and RG of the MRUTS.It only uses the closed-loop I/O data of the MRUTS to predict the system behaviors of the future time,and the practical I/O constraints of the MRUTS are also included for the urban traffic control formulation.