Adaptive Flight Safety Control Of Aircraft With Unknown Faults

Due to the development of aviation science, aircraft have become most commonly used transportation.Thus flight safety is primary concern to aircraft manufacturers, flight crews and passengers. Although modern aircraft have applied a lot of safety measures, including multiple sensor systems and redundant actuators, flight accidents still cannot be avoided. Pilot perception in the cockpit is limited, making it difficult to recognize fault details and keep stable flight control. Flight safety control system can assist the pilot to make right judgment and proper operation, and which is of great significance to improve survivability of aircraft with faults or damages.This dissertation proposes a trim-state-based adaptive flight safety control strategy for damaged aircrafts. Based on a real comprehensive nonlinear aircraft dynamic model, this dissertation studies several key issues of flight control under unknown faults condition, including quick adaptive steady control, trim state command following, on-line system identification and trim state discovery. Main works of this dissertation are as follows:First, the nonlinear dynamic model of twin-engine fixed-wing aircraft is established and linearized model is analysed. Based on NASA Generic Transport Model(GTM), adaptive flight control method for damaged aircraft is studied. A retrospective cost adaptive control(RCAC) method is designed to keep steady flight under several fault conditions, including control surface faults, aerodynamic coefficients change, icing and engine faults. In contrast with traditional faultdiagnosis-based fault tolerant control, RCAC can adaptively change controller gains based on control inputs and system outputs, to keep the damaged aircraft stable using the functional-redundantly actuators, although fault details are unknown.Second, considering the situation where the fault details cannot be estimated accurately, in order to gradually obtain enough trim state data and flight envelope information, on-line retrospective cost system identification(RCSI) method based on local linearized aircraft dynamic model is designed. RCSI can focus on the identified target on a given subsystem and solve the system identification problem, when the inputs and outputs of target subsystem cannot be accurately measured. Considering identification of uncertain parameters in local linearized aircraft dynamic model, many simulation analyses are given, including constant parameter estimate, time-varying parameter estimate, and multiple parameters estimate. Accuracies and speeds of identification under different sensor noises are analyzed. To identify uncertain parameter which appears in multiple locations in state matrix, a Jacobian-based subsystem feedback structure is developed to estimate airspeed variations based on linearized longitudinal aircraft dynamic model. Consider the situation that left wing of aircraft is damaged, based on local linearized dynamic model, all the entries in the state function are accurately identified using RCSI.Third,on the problem that, in some fault situation, trim state database which can satisfy emergency landing trajectory planning need to be built online, a trim state discovery algorithm of damaged aircraft based on artificial potential field is proposed. According to local flight envelope information, exploration planning process based on attractive goal force and obstacle repulsive force is given. To address the problem of converging to local minimum and low efficiency, the three-dimensional flight state searching algorithm is proposed. Simulation results validate the proposed methods. To deal with the feasibility problem of flight trajectories in the trim state discovery process, based on different applications, two flight state space and physical space combined obstacle-avoiding algorithms are proposed. In the first obstacle-avoiding algorithm, each physical constraint is mapped to generate a repulsive force in state space, and the advantage of this algorithm is that the damaged aircraft can keep smooth flight transition and be prevented from fault deterioration. When an aircraft has to avoid physical obstacle during minimal time, a trajectory-prediction-based algorithm is proposed to improve obstacle avoiding efficiency during trim state exploration. The second obstacle-avoiding algorithm is based on trajectory prediction, through this algorithm, aircraft can avoid physical obstacle during minimal time and the obstacle avoiding efficiency is improved. Simulation results validate these two obstacle-avoiding algorithms.Last, retrospective cost adaptive trim command following control method of fault aircraft is studied based on GTM model. The robustness and following accuracy of the controller are evaluated and simulation results validate this method. To address the problem that fault condition may deteriorate, parameters in trim state are extended and an envelope-protection adaptive command following control structure is proposed. Simulation results show that this control structure can effectively enhance flight safety of fault aircraft. And, a whole emergency landing control plan for left wing damaged GTM model is designed, and safe landing for damaged aircraft is achieved.