Fault Detection For Fixed Wing Unmanned Aerial Vehicle With Actuator Saturation

Because of its small size and low cost,the research of UAV technology has become a hot topic.With the UAV increasing flight envelope and worse working environment,the probability of failure of UAV is greatly increased,which will result in huge economic losses.The United States Air Force statistics report said that 78%of the UAV crashed cases were caused by mechanical and electrical faults,however,the replacement cost up to $14 million for a fully equipped military UAV.In order to improve the security and reliability of the system,this thesis research the fault detection problem for UAV flight model.The model-based fault detection approach is a very effective method,but it still needs to be improved.The traditional fault detection method for linear system is difficult to achieve good performance in the system with saturation,and the actual UAV system due to physical limitations will inevitably be saturated.It's very hard to detect a small amplitude of the stuck fault in the existing fault detection methods;The existing adaptive observer can't consider both saturation and disturbance effectively.In view of the above problems,with the aid of the finite frequency combined with the generalized KYP lemma H_/H? method and the adaptive method,the existing fault detection methods are improved based on previous works of others.The main work of this thesis is as follows:Firstly,this thesis introduces the characteristics and applications of the UAV,and analyses the development history and research status of fault detection and saturation treatment for UAV.Then the flight process of UAV is analyzed and modeled to obtained the linear model of UAV.The problem statement and preliminaries about fault detection are also given.Secondly,considering the obtained linear system with actuator saturation,this thesis researches outage fault detection with finite frequency reference inputs.Using the method of estimating attraction domain to solve saturation,and introducing a dynamic feedback controller to the system to stabilize the closed loop system,the closed-loop system can be modeled as faulty models and fault-free models.By designing a filter,the residuals are designed to be sensitive to finite frequency reference inputs for the faulty cases and robust against reference inputs for the fault-free case.Frequency domain inequality problems are converted into linear matrix inequality problems by the help of generalized KYP lemma,and the multi objective optimization problem based on linear matrix inequalities is also presented.A simulation example is given to demonstrate the effectiveness of the proposed algorithm.Finally,fault detection problem for stuck fault is investigated with adaptive observer.The fault detection method based on finite frequency can only detect fault by comparing with the threshold value of fault-free operation,however,fault value can be estimated accurately by using adaptive observer.An improved adaptive observer is designed to estimate the offset fault value under the condition of saturation and to make it have the ability to restrain the disturbance.The validity of the results is verified by the simulation results.