Research On Fault Diagnosis Strategy Of Unmanned Helicopter Sensors

UAV have advantages of vertical take-off and landing,low-speed flight,and fixed-point hovering.In recent years,domestic and foreign scholars have been attracted by them,and more functions have been gradually developed.UAV have got further development in military,agricultural and other engineering fields.As a module for measuring the attitude parameters of an unmanned helicopter,the IMU module plays a key role in the normal flight.In order to ensure that unmanned helicopters can fly normally and steadily,ensuring the accuracy of the data monitored by the sensors of the IMU module is the primary problem,we must detect and diagnose the sensors' state of the module so that the whole system can get fault information timely and deal with it in real time.So the fault diagnosis of the sensor on this module is very necessary.This paper firstly analyzes the possible faults of the sensors in the IMU module of the flight control system,and constructs several mathematical models of common faults.Secondly,according to the study of BP neural network and the operating mechanism of BP neural network,author expounded the model of BP neural network in fault diagnosis.Because BP neural network has the defects of local minimum value,this paper proposes a genetic algorithm based on genetic algorithm.Optimize BP neural network fault diagnosis optimization strategy.The selection of network hidden nodes and the selection of optimization algorithms are verified by experiments.The validation results show that the optimized algorithm is feasible and can greatly improve the convergence speed and accuracy of the network.Then,according to the proposed fault diagnosis and optimization strategy,an optimized neural network observer model is constructed,and a UAV nonlinear dynamics system model is established.The fault diagnosis system is analyzed,and false alarms and omissions of the system are targeted at the same time.We use the sequential probability ratio criterion to judge the system residuals,which greatly improves the accuracy of fault diagnosis and the system's anti-jamming capability.The experiment proves that the entire network is very robust,and this strategy only requires a set of sensors to complete the detection of the system and greatly improves the performance of the entire system.Finally,using MATLAB to create a simple flight control system GUI simulation environment to complete the simulation of the algorithm,according to an existing aircraft model,and the fault diagnosis optimization strategy proposed in this paper,taking the output value of the gyroscope attitude angle as Estimating parameters,constructing corresponding pitch angle observers,roll angle observers,and yaw angle observers,and online estimation of the gyroscope output.Under the normal working conditions of the system,the residual value of the comparison between the estimated output value of the observer and the actual output value of the original system is basically 0.Because the UAV system has strong coupling,when a module fails,the residuals of other modules will also be deviated.The residual is judged to determine whether a fault has occurred.If a fault occurs,the fault point can be quickly located,thereby improving the robustness of the entire system.The simulation results also show that the strategy can be applied to the fault diagnosis of the sensor,as to complete the online detection work and meet the requirements of the system.