Design and implementation of multivariable cooperative control and failure accommodation.

摘要

A mechanical system designed for precise positioning usually consists of many individually actuated subsystems. Redundancy and diversity properties of subsystems are used in the design of a reconfigurable control scheme to achieve optimal coordination and fault accommodation. This dissertation introduces a general procedure of synthesizing multivariable control algorithms that utilize the redundancy and diversity of the subsystems, and discussed the related controller design and implementation issues through three applications with different emphases. A robot target tracking system is employed to demonstrate that the subsystems with different talents work cooperatively towards the same objective. The tracking controller also has the ability to instruct the remaining healthy subsystems to achieve fault accommodation when failure occurs. For each fault scenario only one fixed accommodation controller is needed to address a variety of mechanical and electrical failure conditions. A hard drive dual-stage actuator servo control application illustrates the synthesis of cooperative systems where the participation of each subsystem needs to be adjusted and their limitations accommodated. This dissertation also discusses the implementation of designed digital tracking controllers using embedded microprocessor. A spinning vehicle maneuver control project exemplifies the design process from system modeling to control synthesis and implementation on DSP microprocessor circuits. The discussion on the designed sinusoidal tracking controller's robustness against command frequency variation and other practical issues arising in the implementation are also included.