FAULT TOLERANCE EVALUATION OF A NEURAL PREDICTIVE FLIGHT CONTROL SYSTEM

The nonlinear control of aircraft is now activity sort in an attempt to improve control performance of advanced aircraft. Aircraft control is a formidable problem requiring the controller not only to control some complex dynamics but in the multi-axis cases involving complex coupling. Control can be further complicated if the aircraft is subjected to a subsystem fault, e.g. actuator fault. These faults can lead to catastrophic failures resulting in the loss of an aircraft which should be avoided by all means. This motivates the development of fault tolerant controllers. Fault tolerant controllers offer not only improved safety in the event of a subsystem failure but are expected to reduce both manufacturing and operating costs. This paper will demonstrate the potential and the inherent fault tolerance of a multi-axis DMC-based neural predictive controller in the control of a nonlinear dynamic aircraft model.