Sliding Mode Fault Tolerant Reconfigurable Control against Aircraft Control Surface Failures

Operational failure of control surfaces is one of the main reasons leading to aircraft crash. Since the conventional control methodologies are not adequate to accommodate such failures, fault tolerant control (FTC) is required for safety critical system. The invariance property and unique synthesization procedure of sliding mode control (SMC) make it one of the most competitive candidates for FTC. In this thesis, SMC-based FTC methods for nonlinear systems are developed to handle both partial loss faults and total failures in the control surfaces. The first SMC-based FTC is developed to accommodate both modeling uncertainty and uncertainty incurred by the faults. Different design parameters are utilized to deal with the uncertainty incurred by fault and that due to modeling errors respectively in the SMC design. Direct adaptive control is combined into such a SMC to alleviate the requirement of the a priori knowledge of the uncertainty bounds. The second SMC-based FTC is developed to redistribute the control effort between faulty regular actuator and redundant actuator autonomously based on effectiveness of the regular actuators. The tolerability of the developed controller is characterized by the amount of fault that controller can deal with. It is used as the threshold to activate the redundant actuator when the regular actuator cannot accommodate the fault alone. In order to obtain the effectiveness of the actuator, special sensors or fault detection and diagnosis (FDD) schemes are required. Special sensors are costly and additional design of the system is required. Using FDD, during the period from the moment when fault occurred to that when the effectiveness information can be obtained, the system is under the danger of losing control. The third SMC-based FTC without a dedicated FDD is developed based on the absolute value quantity of switching surface. The control effort is redistributed to regular and redundant actuator autonomously by monitoring the absolute value of the sliding surface. The validity of the proposed algorithms is verified on a high fidelity model of Boeing 747-100/200