Design On UAV Fault Diagnosis System Based On CFT Probabilistic Neural Network

The actuator system of the commercial six-rotor UAV is subject to many factors such as cost and weight,and is prone to various failures.In serious cases,it may even cause a crash.In the case of early minor faults in the drone,if the fault can be diagnosed in a timely and accurate manner,the repair can be checked in advance to avoid major accidents.Therefore,early fault diagnosis is of great significance to the safety of the UAV_In this paper,the six-rotor UAV is taken as the research object.For the single failure mode of the motor and based on the Persistent Excitation(PE)and Conformal Fourier Transform(CFT).The paper mainly designs a persistent excitation-type conformal Fourier transform fault diagnosis method and a probabilistic neural network fault diagnosis model based on PE and CFT,and it realizes the fault diagnosis of the aircraft.The main work of this paper includes the following:1.Establish a six-rotor UAV model based on MATLAB/Simulink.Firstly,the basic structure and flight principle of the six-rotor UAV are introduced,and the UAV modeling work is completed based on the Simulink platform.Then use the wavelet transform to extract the noise from the real flight data and simulate the random interference.Finally,import the both into the model to and the simulation analysis is carried out.2.Design a fault diagnosis method for persistent excitation-type conformal Fourier transform.Firstly,a corresponding fault model is established further for the single failure situation of the UAV motor based on the previous six-rotor Simulink model.Then,the persistent excitation input(PE)and the conformal Fourier transform(CFT)are designed respectively to form a persistent excitation-type conformal Fourier transform fault diagnosis method,and the simulation and verification are completed for the fault diagnosis method by model.3.Design a probabilistic neural network fault diagnosis method based on PE and CFT.Based on persistent excitation-type conformal Fourier transform method,the design and establishment of the probabilistic neural network are carried out,and the feasibility of the network model is analyzed by simulation data.The good pattern recognition ability of the model is verified,and the fault classification of the motor is realized.4.Design a real UAV fault diagnosis system based on PE and CFT.Firstly,the real UAV database of the Pixhawk system is established,and realize the simulation of the aircraft motor failure.Then,the flight test of the UAV is designed,and the preprocessing of the flight data samples is completed.Finally,design the probabilistic neural network model in the MATLAB environment,and the fault mode classification and verification of the actual flight data is carried out.In summary,this paper proposes two new fault diagnosis methods,and develops a corresponding real UAV fault diagnosis system.Through simulation research and actual flight test,it is verified that the two fault diagnosis methods proposed in this paper have higher fault diagnosis accuracy and have certain application value.