Research On Data-driven Fault-tolerant Control Methods For Unmanned Aerial Vehicle

With the advantages of fast flight speed,high cruising altitude and strong maneuverability,fixed-wing UAVs can quickly reach the near-Earth space and also reach any designated location within a short period of time,representing the future development trend of near-Earth space vehicles and occupying an important position in the military and civilian fields,with great development potential and application value.The increasingly widespread application of fixed-wing UAVs has triggered users to think about their flight safety,especially in the case of failure of UAV body parts,and how to achieve fault-tolerant control of the UAV system in the case of failure of the body parts to ensure that the UAV still has a certain degree of flight stability is of great significance.Given the complexity of the fixed-wing UAV structure,the modeling process is very complicated and the accuracy is not guaranteed,but with the development of advanced sensor technology,the data during the flight of the UAV can be effectively stored and used,which contains the characteristics and laws of the system and provides the necessary basis for the data-driven fault-tolerant control method.In this paper,firstly,the system model,mutual quality decomposition technique and Youla parameterization theory are studied,the concepts of fitness and internal stability are introduced,and the integration method is obtained by analyzing and summarizing on the basis of the above theories.This method is able to generate compensating control signals to suppress collective faults and achieve fault-tolerant control of the system without changing the original controller of the system.Next,the data-driven residual generation method is introduced,and the data-driven form is elicited through the model approach to obtain a stable kernel description of the data-driven form of the system.Next,the stability of the integration framework is analyzed,and the gradient update method of the Youla parameterization matrix is given.Then secondly,the fault-tolerant control method for sensor failure is introduced,and the core of this method is reconstructing the sensor signal,and two methods are given from different perspectives.The article concludes with a simulation example validation of the theoretical content and a summary of the experimental methods and results.