Separate And Embedded Fault Diagnosis And Fault Tolerance Research Of UAV Flight Control System

The UAV is a non-manned aircraft operated by radio remote sensing equipment and self-provided program control devices,and the flight control system is the core of the autonomous flight of the UAV.Therefore,the reliability and stability of the flight control system are particularly important.At present,there are still many problems.For example,the analysis methods and mining methods for massive data are poor,and the accurate UAV fault-tolerant control model is difficult to establish.The separate fault diagnosis algorithm can identify the fault information offline and provide a reliable basis for the maintenance and management of the drone system.The embedded fault diagnosis and fault-tolerant technology can realize online adjustment and improve the reliability index of the drone system.Therefore,discrete and embedded fault diagnosis and fault tolerance technology is an important research hotspot in the field of UAV technology.Firstly,the signal is decomposed by the variational mode decomposition method,and then the multi-scale reconstruction index and continuous wavelet transform method are used to obtain the time-domain characteristics and time-frequency characteristics of the reconstructed signal.In view of the accuracy and rapidity of convolutional neural network(CNN)in mining effective information of massive data,this paper uses CNN algorithm to perform separate fault diagnosis research on time domain and time-frequency characteristics.The simulation results show that the CNN diagnosis algorithm based on time domain features consumes less time,while the time-frequency feature based CNN algorithm has higher diagnostic accuracy.Combining the advantages of both,you can build an integrated algorithm with relatively short time and higher precision.Secondly,based on the single-sample time domain and time-frequency feature CNN algorithm,in order to fuse multiple single-sample features,this paper studies a seperate fault diagnosis integrated algorithm based on multi-sample features.The self-sampling method is used to fuse the time domain and time-frequency features,then the CNN algorithm is used as the base classifier,and the majority voting method is used to combine the results of the CNN algorithm.The simulation results show that the average diagnostic accuracy is at least 99.89%.Compared with the single-sample feature CNN algorithm,the integrated algorithm has higher test accuracy and lower generalization error.Thirdly,in order to simplify the controller structure and deal with the fault information that is difficult to model,this paper adopts the adaptive terminal sliding mode control method to design a height and attitude fault-tolerant controller for the UAV flight control model.The simulation shows that the position tracking error range is [-0.2,0.2].Fourth,in order to realize the automatic fault detection and fault tolerance of the UAV system,this paper combines the embedded fault diagnosis method with fault tolerance.The expansion observer is used to estimate the fault,realize online diagnosis and reconstruct the fault signal,and then reconstruct the sliding mode fault-tolerant controller according to the fault information to compensate the fault.The simulation shows that the error estimation error range is [-0.1,0.1],the position tracking error range is [0,0.04 ],and the angle tracking error converges to zero.Finally,this paper studies the effective fault diagnosis algorithm and fault-tolerant control method,and provides a reliable basis for establishing the health management strategy of the UAV fault device and adjusting the fault-tolerant controller of the UAV system.