Fault tolerant control design for parameter dependent systems

Abstract

This paper addresses application of a fault tolerant control design technique for parameter dependent systems in two phases, namely (1) control law design and (2) sensor and actuator failure detection and isolation and reconfiguring. Traditional solutions (i.e. gain scheduling) to the parameter dependent control problem typically assume slow variation of the system dynamics, which is not necessarily valid for the case of fast parameter variations. The basis of the technique is the casting of the problem in a framework, which is an extension of feedback linearization, to account for the time varying nature of the system. A coordinates transformation, when combined with a linearizing control law including terms accounting for time variation of the parameters, defines a structure for designing the control system for parameter dependent systems with fast parameter variation. The framework is extended to the case of control system failure detection and isolation (FDI) and reconfiguring, thus facilitating design of sensor and actuator failure detection algorithms for parameter dependent systems. The combined control and FDI systems then offer a structured design approach for fault tolerant control of parameter dependent systems. The techniques are demonstrated on (1) a unicycle robot, (2) a Perseus UAV, and (3) a Cessna O2 aircraft.