Date of Award

Fall 1997

Document Type


Degree Name

Doctor of Philosophy in Electrical Engineering - (Ph.D.)


Electrical and Computer Engineering

First Advisor

Timothy Nam Chang

Second Advisor

Bernard Friedland

Third Advisor

Andrew Ulrich Meyer

Fourth Advisor

Marshall Chuan Yung Kuo

Fifth Advisor

Edwin Hou

Sixth Advisor

I-Jy Steven Chien


Reliable control concerns the ability of closed loop system to maintain stability and regulation properties during arbitrary sensor, controller, and actuator failure. Reliable control research has been an active research topic for more than 10 years.

Recent approach for reliable control includes the H method, the algebraic factorization design, and the robust servomechanism control. These methods have been surveyed and discussed in this thesis with the robust servomechanism control methodology serving as the basis of the research development of this work.

In this thesis, the reliable control for large-scale, multi-input/output linear system is considered. Two concepts of reliable control are introduced in this work: (1) Decentralized Robust Servomechanism Problem with Complete Reliability (DRSPwCR) and (2) Block Decentralized Robust Servo Problem with Complete Reliability (BDRSPwCR). The DRSPwCR solves the reliable control problem by applying strict diagonal decentralized controller configurations. The BDRSPwCR solves the reliable control problem by applying block diagonal decentralized controller configurations.

Research results of solving DRSPwCR for the class of minimum phase systems is first developed in this work. The problem is solved by applying strict decentralized PIDr control to an otherwise unreliable plant and thus significantly extending the class of processes that can be controlled reliably. Research results of solving BDRSPwCR is developed for plants which have a pre-imposed block diagonal structure or plants with non-minimum phase minors. The reliable control conditions for an arbitrary linear system is then analyzed, and a general controller synthesis for solving the reliable control problem for arbitrary linear system is given in this work.

The DRSPwCR can be applied in many industry areas as well as in the transportation area. In this work, the reliable control results are applied in the urban vehicle traffic network. A traffic queue length model is developed, a control algorithm is synthesized, and simulations are made under different traffic subsystem failure modes such as non-functioning traffic lights, traffic accidents, and intersection blockage, etc.

Finally, future research topics such as to relax the constraints of plants to achieve reliable control and to optimize the closed loop system dynamic performances, etc. are proposed.