AbstractsEngineering

Distributed Fault Diagnosis of Interconnected Nonlinear Uncertain Systems

by Qi Zhang




Institution: Wright State University
Department: Engineering PhD
Degree: PhD
Year: 2013
Keywords: Electrical Engineering; fault diagnosis; interconnected systems; adaptive learning
Record ID: 2002001
Full text PDF: http://rave.ohiolink.edu/etdc/view?acc_num=wright1377888655


Abstract

Fault diagnosis is crucial in achieving safe and reliable operations of interconnected controlsystems. This dissertation presents a distributed fault detection and isolation (FDI)method for interconnected nonlinear uncertain systems. The contributions of this dissertationinclude the following: First, the detection and isolation problem of process faults in aclass of interconnected input-output nonlinear uncertain systems is investigated. A novelfault detection and isolation scheme is devised, and the fault detectability and isolabilityconditions are rigorously investigated, characterizing the class of faults in each subsystemthat are detectable and isolable by the proposed distributed FDI method. Second, a distributedsensor fault FDI scheme is developed in a class of interconnected input-outputnonlinear systems where only the measurable part of state variables are directly affected bythe interconnections between subsystems. A class of multimachine power systems is usedas an application example to illustrate the effectiveness of the proposed approach. Third,the previous results are extended to a class of interconnected input-output nonlinear systemswhere both the unknown and the measurable part of system states of each subsystemare directly affected by the interconnections between subsystems. In this case, the faultpropagation effect among subsystems directly affects the unknown part of state variables ofeach subsystem. Thus, the problem considered is more challenging than what is describedabove. Finally, a fault detection scheme is presented for a more general distributed nonlinearsystems. With a removal of a restrictive limitation on the system model structure,the results described above are extended to a class of interconnected nonlinear uncertainsystems with a more general structure.In addition, the effectiveness of the above fault diagnosis schemes is illustrated by usingsimulations of interconnected inverted pendulums mounted on carts and multi-machinepower systems. Different fault scenarios are considered to verify the diagnosis performances, and the satisfactory performances of the proposed diagnosis scheme are validated by thegood simulation results. Some interesting future research work is also discussed.