Research On Fault Diagnosis And Fault Tolerant Control Of Aircraft Attitude Control System With Actuator Faults

An aircraft is usually composed of many subsystems,and the attitude control system is the most critical one.Because of its long-term complex and changeable flight environment,it is prone to failure.Once failure occurs,if no measures are taken to deal with it,it will cause immeasurable losses.Therefore,it is of great significance to carry out fault diagnosis and fault-tolerant control technology research on attitude control system.Aiming at the problem of the actuator failure of the aircraft attitude control system,this paper studies the fault detection,fault diagnosis and fault-tolerant control methods.The main work is as follows:For the general aircraft attitude control system,in order to facilitate the development of subsequent research,the basic parameters of the aircraft and the definition and conversion of the coordinate system are first defined.Then four kinds of posture description methods are analyzed,and kinematics equations and dynamics equations are derived on this basis.Finally,the common failures of actuators are analyzed,and the mathematical expressions of failures are listed.Aiming at the additive faults and multiplicative faults that are most likely to occur in actuator faults,two model-based fault detection methods are studied.The effects of two types of faults and their magnitudes on the attitude control system of the aircraft are analyzed by means of numerical simulation,and the necessity of the next step of fault diagnosis is explained.Then the fault detection method based on the upper bound of the disturbance is studied,as long as the residual is detected After exceeding the threshold related to interference,it is judged that the fault has occurred.Finally,a robust observer with fewer parameters is designed for fault detection.Compared with the former,it does not require information such as the interference upper bound.The results show that both methods can be accurate and effective.If failures are detected locally,the latter is more advantageous in terms of timeliness of detection.The data-driven method is used to further diagnose the size of the fault.On the basis of the posture simulation in the previous chapter,the experimental data set is collected as input data,and the network output is the fault size.In order to make the network pay more attention to the fault characteristic information,a diagnosis network was built by introducing an attention mechanism based on the one-dimensional convolutional neural network and the long-and short-term memory network.In order to achieve the purpose of high-precision diagnosis of the size of the fault,simulation experiments were carried out on the network hyperparameters and the diagnosis effect,and a set of parameter combinations were selected through the simulation test.In order to further verify the effectiveness of this data-driven method in the field of actuator fault diagnosis,the diagnosis network is compared and simulated with a series network composed of a one-dimensional convolutional neural network and a long and short-term memory network.According to the detection and diagnosis results of the first two chapters,in order to ensure that the attitude control system can maintain stability and have good dynamic performance after the fault occurs,a fault-tolerant controller based on the backstepping method under the preset performance constraints is designed and verified.Firstly,using the principle of preset performance control,the attitude error performance constraint problem in this paper is transformed into a bounded control problem,and the inequality constraint becomes a convenient equality constraint.Then,the converted system is combined with the backstepping method to design a fault-tolerant controller and the Lyapunov function is used to strictly prove the stability,ensuring that the system remains stable after a fault occurs and the transient performance is maintained within the preset performance constraints.